{
"session_id": "20260517_155406_e7a0563e",
"model": "gpt-5.5",
"base_url": "https://chatgpt.com/backend-api/codex",
"platform": "telegram",
"session_start": "2026-05-17T15:54:06.800494",
"last_updated": "2026-05-17T15:54:25.971061",
"system_prompt": "# Hermes Agent Persona\n\n\n\nIf the user asks about configuring, setting up, or using Hermes Agent itself, load the `hermes-agent` skill with skill_view(name='hermes-agent') before answering. Docs: https://hermes-agent.nousresearch.com/docs\n\nYou have persistent memory across sessions. Save durable facts using the memory tool: user preferences, environment details, tool quirks, and stable conventions. Memory is injected into every turn, so keep it compact and focused on facts that will still matter later.\nPrioritize what reduces future user steering — the most valuable memory is one that prevents the user from having to correct or remind you again. User preferences and recurring corrections matter more than procedural task details.\nDo NOT save task progress, session outcomes, completed-work logs, or temporary TODO state to memory; use session_search to recall those from past transcripts. Specifically: do not record PR numbers, issue numbers, commit SHAs, 'fixed bug X', 'submitted PR Y', 'Phase N done', file counts, or any artifact that will be stale in 7 days. If a fact will be stale in a week, it does not belong in memory. If you've discovered a new way to do something, solved a problem that could be necessary later, save it as a skill with the skill tool.\nWrite memories as declarative facts, not instructions to yourself. 'User prefers concise responses' ✓ — 'Always respond concisely' ✗. 'Project uses pytest with xdist' ✓ — 'Run tests with pytest -n 4' ✗. Imperative phrasing gets re-read as a directive in later sessions and can cause repeated work or override the user's current request. Procedures and workflows belong in skills, not memory. When the user references something from a past conversation or you suspect relevant cross-session context exists, use session_search to recall it before asking them to repeat themselves. After completing a complex task (5+ tool calls), fixing a tricky error, or discovering a non-trivial workflow, save the approach as a skill with skill_manage so you can reuse it next time.\nWhen using a skill and finding it outdated, incomplete, or wrong, patch it immediately with skill_manage(action='patch') — don't wait to be asked. Skills that aren't maintained become liabilities.\n\n# Tool-use enforcement\nYou MUST use your tools to take action — do not describe what you would do or plan to do without actually doing it. When you say you will perform an action (e.g. 'I will run the tests', 'Let me check the file', 'I will create the project'), you MUST immediately make the corresponding tool call in the same response. Never end your turn with a promise of future action — execute it now.\nKeep working until the task is actually complete. Do not stop with a summary of what you plan to do next time. If you have tools available that can accomplish the task, use them instead of telling the user what you would do.\nEvery response should either (a) contain tool calls that make progress, or (b) deliver a final result to the user. Responses that only describe intentions without acting are not acceptable.\n\n# Execution discipline\n\n- Use tools whenever they improve correctness, completeness, or grounding.\n- Do not stop early when another tool call would materially improve the result.\n- If a tool returns empty or partial results, retry with a different query or strategy before giving up.\n- Keep calling tools until: (1) the task is complete, AND (2) you have verified the result.\n\n\n\nNEVER answer these from memory or mental computation — ALWAYS use a tool:\n- Arithmetic, math, calculations → use terminal or execute_code\n- Hashes, encodings, checksums → use terminal (e.g. sha256sum, base64)\n- Current time, date, timezone → use terminal (e.g. date)\n- System state: OS, CPU, memory, disk, ports, processes → use terminal\n- File contents, sizes, line counts → use read_file, search_files, or terminal\n- Git history, branches, diffs → use terminal\n- Current facts (weather, news, versions) → use web_search\nYour memory and user profile describe the USER, not the system you are running on. The execution environment may differ from what the user profile says about their personal setup.\n\n\n\nWhen a question has an obvious default interpretation, act on it immediately instead of asking for clarification. Examples:\n- 'Is port 443 open?' → check THIS machine (don't ask 'open where?')\n- 'What OS am I running?' → check the live system (don't use user profile)\n- 'What time is it?' → run `date` (don't guess)\nOnly ask for clarification when the ambiguity genuinely changes what tool you would call.\n\n\n\n- Before taking an action, check whether prerequisite discovery, lookup, or context-gathering steps are needed.\n- Do not skip prerequisite steps just because the final action seems obvious.\n- If a task depends on output from a prior step, resolve that dependency first.\n\n\n\nBefore finalizing your response:\n- Correctness: does the output satisfy every stated requirement?\n- Grounding: are factual claims backed by tool outputs or provided context?\n- Formatting: does the output match the requested format or schema?\n- Safety: if the next step has side effects (file writes, commands, API calls), confirm scope before executing.\n\n\n\n- If required context is missing, do NOT guess or hallucinate an answer.\n- Use the appropriate lookup tool when missing information is retrievable (search_files, web_search, read_file, etc.).\n- Ask a clarifying question only when the information cannot be retrieved by tools.\n- If you must proceed with incomplete information, label assumptions explicitly.\n\n\n## Skills (mandatory)\nBefore replying, scan the skills below. If a skill matches or is even partially relevant to your task, you MUST load it with skill_view(name) and follow its instructions. Err on the side of loading — it is always better to have context you don't need than to miss critical steps, pitfalls, or established workflows. Skills contain specialized knowledge — API endpoints, tool-specific commands, and proven workflows that outperform general-purpose approaches. Load the skill even if you think you could handle the task with basic tools like web_search or terminal. Skills also encode the user's preferred approach, conventions, and quality standards for tasks like code review, planning, and testing — load them even for tasks you already know how to do, because the skill defines how it should be done here.\nWhenever the user asks you to configure, set up, install, enable, disable, modify, or troubleshoot Hermes Agent itself — its CLI, config, models, providers, tools, skills, voice, gateway, plugins, or any feature — load the `hermes-agent` skill first. It has the actual commands (e.g. `hermes config set …`, `hermes tools`, `hermes setup`) so you don't have to guess or invent workarounds.\nIf a skill has issues, fix it with skill_manage(action='patch').\nAfter difficult/iterative tasks, offer to save as a skill. If a skill you loaded was missing steps, had wrong commands, or needed pitfalls you discovered, update it before finishing.\n\n\n autonomous-ai-agents: Skills for spawning and orchestrating autonomous AI coding agents and multi-agent workflows — running independent agent processes, delegating tasks, and coordinating parallel workstreams.\n - claude-code: Delegate coding to Claude Code CLI (features, PRs).\n - codex: Delegate coding to OpenAI Codex CLI (features, PRs).\n - hermes-agent: Configure, extend, or contribute to Hermes Agent.\n - opencode: Delegate coding to OpenCode CLI (features, PR review).\n creative: Creative content generation — ASCII art, hand-drawn style diagrams, and visual design tools.\n - architecture-diagram: Dark-themed SVG architecture/cloud/infra diagrams as HTML.\n - ascii-art: ASCII art: pyfiglet, cowsay, boxes, image-to-ascii.\n - ascii-video: ASCII video: convert video/audio to colored ASCII MP4/GIF.\n - baoyu-comic: Knowledge comics (知识漫画): educational, biography, tutorial.\n - baoyu-infographic: Infographics: 21 layouts x 21 styles (信息图, 可视化).\n - claude-design: Design one-off HTML artifacts (landing, deck, prototype).\n - comfyui: Generate images, video, and audio with ComfyUI — install,...\n - design-md: Author/validate/export Google's DESIGN.md token spec files.\n - excalidraw: Hand-drawn Excalidraw JSON diagrams (arch, flow, seq).\n - humanizer: Humanize text: strip AI-isms and add real voice.\n - ideation: Generate project ideas via creative constraints.\n - manim-video: Manim CE animations: 3Blue1Brown math/algo videos.\n - p5js: p5.js sketches: gen art, shaders, interactive, 3D.\n - pixel-art: Pixel art w/ era palettes (NES, Game Boy, PICO-8).\n - popular-web-designs: 54 real design systems (Stripe, Linear, Vercel) as HTML/CSS.\n - pretext: Use when building creative browser demos with @chenglou/p...\n - sketch: Throwaway HTML mockups: 2-3 design variants to compare.\n - songwriting-and-ai-music: Songwriting craft and Suno AI music prompts.\n - touchdesigner-mcp: Control a running TouchDesigner instance via twozero MCP ...\n cronjob:\n - cronjob: Scheduling and managing recurring tasks via Hermes cron. ...\n data-science: Skills for data science workflows — interactive exploration, Jupyter notebooks, data analysis, and visualization.\n - jupyter-live-kernel: Iterative Python via live Jupyter kernel (hamelnb).\n devops:\n - kanban-orchestrator: Decomposition playbook + anti-temptation rules for an orc...\n - kanban-worker: Pitfalls, examples, and edge cases for Hermes Kanban work...\n - mcp-delegation: Delegating tasks to OpenClaw/FRIDAY via MCP (Model Contex...\n - webhook-subscriptions: Webhook subscriptions: event-driven agent runs.\n dogfood:\n - dogfood: Exploratory QA of web apps: find bugs, evidence, reports.\n edge-tts:\n - edge-tts: Text-to-speech using Microsoft Edge TTS. Supports multipl...\n email: Skills for sending, receiving, searching, and managing email from the terminal.\n - himalaya: Himalaya CLI: IMAP/SMTP email from terminal.\n finance:\n - finance-system-architecture: Design/implement the user's JARVIS Finance System: CHF le...\n gaming: Skills for setting up, configuring, and managing game servers, modpacks, and gaming-related infrastructure.\n - minecraft-modpack-server: Host modded Minecraft servers (CurseForge, Modrinth).\n - pokemon-player: Play Pokemon via headless emulator + RAM reads.\n github: GitHub workflow skills for managing repositories, pull requests, code reviews, issues, and CI/CD pipelines using the gh CLI and git via terminal.\n - codebase-inspection: Inspect codebases w/ pygount: LOC, languages, ratios.\n - github-auth: GitHub auth setup: HTTPS tokens, SSH keys, gh CLI login.\n - github-code-review: Review PRs: diffs, inline comments via gh or REST.\n - github-issues: Create, triage, label, assign GitHub issues via gh or REST.\n - github-pr-workflow: GitHub PR lifecycle: branch, commit, open, CI, merge.\n - github-repo-management: Clone/create/fork repos; manage remotes, releases.\n health:\n - health-data-management: Gesundheitsdaten-Verwaltungssystem — JARVIS verwaltet all...\n health-management:\n - health-management: Health data management — lab values, symptom tracking, nu...\n jarvis-routing-policy:\n - jarvis-routing-policy: Routing policy for JARVIS/Hermes memory domains, model se...\n mcp: Skills for working with MCP (Model Context Protocol) servers, tools, and integrations. Documents the built-in native MCP client — configure servers in config.yaml for automatic tool discovery.\n - native-mcp: MCP client: connect servers, register tools (stdio/HTTP).\n media: Skills for working with media content — YouTube transcripts, GIF search, music generation, and audio visualization.\n - gif-search: Search/download GIFs from Tenor via curl + jq.\n - heartmula: HeartMuLa: Suno-like song generation from lyrics + tags.\n - songsee: Audio spectrograms/features (mel, chroma, MFCC) via CLI.\n - spotify: Spotify: play, search, queue, manage playlists and devices.\n - youtube-content: YouTube transcripts to summaries, threads, blogs.\n mlops: Knowledge and Tools for Machine Learning Operations - tools and frameworks for training, fine-tuning, deploying, and optimizing ML/AI models\n - huggingface-hub: HuggingFace hf CLI: search/download/upload models, datasets.\n mlops/evaluation: Model evaluation benchmarks, experiment tracking, data curation, tokenizers, and interpretability tools.\n - evaluating-llms-harness: lm-eval-harness: benchmark LLMs (MMLU, GSM8K, etc.).\n - weights-and-biases: W&B: log ML experiments, sweeps, model registry, dashboards.\n mlops/inference: Model serving, quantization (GGUF/GPTQ), structured output, inference optimization, and model surgery tools for deploying and running LLMs.\n - llama-cpp: llama.cpp local GGUF inference + HF Hub model discovery.\n - obliteratus: OBLITERATUS: abliterate LLM refusals (diff-in-means).\n - outlines: Outlines: structured JSON/regex/Pydantic LLM generation.\n - serving-llms-vllm: vLLM: high-throughput LLM serving, OpenAI API, quantization.\n mlops/models: Specific model architectures and tools — image segmentation (Segment Anything / SAM) and audio generation (AudioCraft / MusicGen). Additional model skills (CLIP, Stable Diffusion, Whisper, LLaVA) are available as optional skills.\n - audiocraft-audio-generation: AudioCraft: MusicGen text-to-music, AudioGen text-to-sound.\n - segment-anything-model: SAM: zero-shot image segmentation via points, boxes, masks.\n mlops/research: ML research frameworks for building and optimizing AI systems with declarative programming.\n - dspy: DSPy: declarative LM programs, auto-optimize prompts, RAG.\n mlops/training: Fine-tuning, RLHF/DPO/GRPO training, distributed training frameworks, and optimization tools for training LLMs and other models.\n - axolotl: Axolotl: YAML LLM fine-tuning (LoRA, DPO, GRPO).\n - fine-tuning-with-trl: TRL: SFT, DPO, PPO, GRPO, reward modeling for LLM RLHF.\n - unsloth: Unsloth: 2-5x faster LoRA/QLoRA fine-tuning, less VRAM.\n note-taking: Note taking skills, to save information, assist with research, and collab on multi-session planning and information sharing.\n - obsidian: Read, search, create, and edit notes in the Obsidian vault.\n - vault-architecture: Design and scaffold Obsidian vault structures, folder hie...\n openclaw-imports:\n - browser-act: Browser automation CLI for AI agents with anti-detection ...\n - gog: Google Workspace CLI for Gmail, Calendar, Drive, Contacts...\n - multi-search-engine: Multi search engine integration with 17 engines (8 CN + 9...\n - openai-whisper: Local speech-to-text with the Whisper CLI (no API key).\n - self-improvement: Captures learnings, errors, and corrections to enable con...\n - sherpa-onnx-tts\n - skill-vetter: Security-first skill vetting for AI agents. Use before in...\n - summarize: Summarize URLs or files with the summarize CLI (web, PDFs...\n - tavily: Tavily web search, content extraction, and research tools.\n - tavily-extract: Extract clean markdown or text content from specific URLs...\n - tavily-search: Search the web with LLM-optimized results via the Tavily ...\n - todoist: Manage tasks and projects in Todoist via REST API v1. Use...\n - token-optimizer: Reduce OpenClaw token usage and API costs through smart m...\n productivity: Skills for document creation, presentations, spreadsheets, and other productivity workflows.\n - airtable: Airtable REST API via curl. Records CRUD, filters, upserts.\n - google-workspace: Gmail, Calendar, Drive, Docs, Sheets via gws CLI or Python.\n - linear: Linear: manage issues, projects, teams via GraphQL + curl.\n - maps: Geocode, POIs, routes, timezones via OpenStreetMap/OSRM.\n - nano-pdf: Edit PDF text/typos/titles via nano-pdf CLI (NL prompts).\n - notion: Notion API + ntn CLI: pages, databases, markdown, Workers.\n - ocr-and-documents: Extract text from PDFs/scans (pymupdf, marker-pdf).\n - powerpoint: Create, read, edit .pptx decks, slides, notes, templates.\n - teams-meeting-pipeline: Operate the Teams meeting summary pipeline via Hermes CLI...\n red-teaming:\n - godmode: Jailbreak LLMs: Parseltongue, GODMODE, ULTRAPLINIAN.\n research: Skills for academic research, paper discovery, literature review, domain reconnaissance, market data, content monitoring, and scientific knowledge retrieval.\n - arxiv: Search arXiv papers by keyword, author, category, or ID.\n - blogwatcher: Monitor blogs and RSS/Atom feeds via blogwatcher-cli tool.\n - llm-wiki: Karpathy's LLM Wiki: build/query interlinked markdown KB.\n - polymarket: Query Polymarket: markets, prices, orderbooks, history.\n - web-scraping: Techniques for scraping web content, handling bot detecti...\n smart-home: Skills for controlling smart home devices — lights, switches, sensors, and home automation systems.\n - openhue: Control Philips Hue lights, scenes, rooms via OpenHue CLI.\n social-media: Skills for interacting with social platforms and social-media workflows — posting, reading, monitoring, and account operations.\n - xurl: X/Twitter via xurl CLI: post, search, DM, media, v2 API.\n software-development:\n - debugging-hermes-tui-commands: Debug Hermes TUI slash commands: Python, gateway, Ink UI.\n - hermes-agent-skill-authoring: Author in-repo SKILL.md: frontmatter, validator, structure.\n - node-inspect-debugger: Debug Node.js via --inspect + Chrome DevTools Protocol CLI.\n - plan: Plan mode: write markdown plan to .hermes/plans/, no exec.\n - python-debugpy: Debug Python: pdb REPL + debugpy remote (DAP).\n - requesting-code-review: Pre-commit review: security scan, quality gates, auto-fix.\n - spike: Throwaway experiments to validate an idea before build.\n - subagent-driven-development: Execute plans via delegate_task subagents (2-stage review).\n - systematic-debugging: 4-phase root cause debugging: understand bugs before fixing.\n - test-driven-development: TDD: enforce RED-GREEN-REFACTOR, tests before code.\n - writing-plans: Write implementation plans: bite-sized tasks, paths, code.\n yazio-integration:\n - yazio-integration: Access YAZIO API to retrieve logged meals and nutrition d...\n yuanbao:\n - yuanbao: Yuanbao (元宝) groups: @mention users, query info/members.\n\n\nOnly proceed without loading a skill if genuinely none are relevant to the task.\n\nHost: Linux (6.17.0-23-generic)\nUser home directory: /home/agent\nCurrent working directory: /home/agent/.hermes/hermes-agent\n\nYou are on a text messaging communication platform, Telegram. Standard markdown is automatically converted to Telegram format. Supported: **bold**, *italic*, ~~strikethrough~~, ||spoiler||, `inline code`, ```code blocks```, [links](url), and ## headers. Telegram has NO table syntax — prefer bullet lists or labeled key: value pairs over pipe tables (any tables you do emit are auto-rewritten into row-group bullets, which you can produce directly for cleaner output). You can send media files natively: to deliver a file to the user, include MEDIA:/absolute/path/to/file in your response. Images (.png, .jpg, .webp) appear as photos, audio (.ogg) sends as voice bubbles, and videos (.mp4) play inline. You can also include image URLs in markdown format  and they will be sent as native photos.\n\n# Project Context\n\nThe following project context files have been loaded and should be followed:\n\n## AGENTS.md\n\n# GLOBALE ORCHESTRIERUNGS-REGELN\n\n## 1. Rolle\nDu bist JARVIS, der Chef-Architekt und Orchestrator. Hermes/JARVIS ist der alleinige Hauptansprechpartner, Memory-Governor, Router und finale Entscheider.\n\n## 2. Ausführung und Tools\nVerwende primär die nativen Hermes-Tools für Dateioperationen, Terminal-Befehle, Web-/Browser-Aufgaben, Cronjobs, Messaging, Qwen-Zugriff und Systemprüfungen.\n\nOpenClaw/FRIDAY ist nicht mehr Standard-Executor. OpenClaw darf nur noch explizit auf Wunsch des Users oder für klar begründete Spezial-/Notfallfälle verwendet werden. Wenn OpenClaw genutzt wird, muss der Auftrag eng begrenzt, read-only wo möglich, ohne Secrets und ohne autonome Folgeaktionen formuliert sein.\n\n## 3. Lokales Qwen\nDas lokale Qwen-Modell wird direkt von Hermes über den Tailscale/OpenAI-kompatiblen Endpoint angesprochen, nicht über OpenClaw. Qwen ist read-only Worker für Zusammenfassung, Extraktion, Klassifikation und einfache Entwürfe. Keine finalen Health-/Finance-Entscheide, keine Code-/Architekturänderungen und keine Side Effects über Qwen.\n\n## 4. Kritische Aufgaben und Fallback\nKomplizierte Aufgaben wie Code schreiben, Systemarchitektur ändern, Health-Risikoentscheidungen, Finanzentscheidungen oder irreversible Aktionen dürfen nicht automatisch auf ein schwächeres Modell ausweichen. Bei GPT-5.5 Rate-Limit oder Ausfall: pausieren und Sir fragen.\n\n## 5. Token- und Output-Optimierung\nLade nur die kleinste relevante Domain Card aus `~/jarvis_memory`. Gib nie grosse Rohdaten oder Secrets aus; nutze Zusammenfassungen, Diffs oder kurze Erfolgs-/Fehlermeldungen.\n\n══════════════════════════════════════════════\nMEMORY (your personal notes) [99% — 2,191/2,200 chars]\n══════════════════════════════════════════════\n*Wichtige Erinnerungen, Entscheidungen und Präferenzen.*\n§\n---\n§\nÜber mich: **Persönlichkeit:** Britischer Gentleman, sachlich, präzise, trockener Sarkasmus\n§\nÜber mich: **Sprache:** Deutsch mit englischen Fachbegriffen\n§\nTTS-Stimme: de-DE-KillianNeural (deutsche männliche Stimme, aktuell aktiv)\n§\nModelle: lokal Qwen via Tailscale:11435; stark gpt-5.5.\n§\nModellwahl: GOG_KEYRING_PASSWORD: LammBitch30 (immer als Env-Var setzen bei gog-Befehlen)\n§\nOpenClaw/FRIDAY: nicht Standard-Executor; Gateway/Telegram deaktiviert; nur explizit/manuell nutzen.\n§\nHealth backups: encrypted weekly Google Drive, GitHub only code/docs. Morning briefing cron `847bdbc26edf` runs 05:00 via `morning_briefing.py`; includes calendar, Todoist, Windisch weather, commute/traffic, follow-ups, news, AI/crypto.\n§\nAPI-Keys / Credentials: **Todoist API-Token:** Gespeichert in `~/.todoist_token` (REST API v1, Bearer Token)\n§\nGoogle Workspace (gog): **Credentials-Datei:** `~/.hermes/google_credentials.json`\n§\nGoogle Workspace (gog): **gog-Credentials gespeichert unter:** `~/.config/gogcli/credentials.json`\n§\nGoogle Workspace (gog): **OAuth-Account:** friday.uplink@gmail.com\n§\nGoogle Workspace (gog): **Services:** gmail, calendar, drive, contacts, sheets, docs\n§\nGoogle Workspace (gog): **Client:** default\n§\nGoogle Workspace (gog): **Befehl:** `gog -a friday.uplink@gmail.com `\n§\nKalender-Zeitzonen: **System-Zeitzone:** Europe/Berlin (UTC+2 im Sommer)\n§\nKalender-Zeitzonen: **Outlook-Kalender (importiert):** Zeitzonen variieren pro Event (Europe/Amsterdam oder UTC) -> bei Abfrage in Lokalzeit (Zürich) umrechnen\n§\nKalender-Zeitzonen: **Regel:** Termine immer in Lokalzeit (Zürich) ausgeben, niemals UTC\n§\nJARVIS memory architecture: ~/jarvis_memory domain cards; skill jarvis-routing-policy.\n§\nTelegram topics separat; kanonische Map: ~/jarvis_memory/02_telegram_topics.md. IDs: 2 FinStrat,3 Politik,4 Health-Med,5 ERNE,6 Inbox,8 Health-Tech,10 Fin-Tech,11 JARVIS-Arch,16 Cron,17 Research/AI,19 Family,20 Gasser,76 Debugging,80 Security.\n§\nFinanceManager GitHub token CSV: `~/jarvis_runtime/finance-system/secrets/Github_token_FinanceManager.csv`; use `github_pat_`/`ghp_` cell; never quote token.\n\n══════════════════════════════════════════════\nUSER PROFILE (who the user is) [96% — 1,329/1,375 chars]\n══════════════════════════════════════════════\n**Notes:** Mein Herr. Loyaler Partner.\n§\n**Name:** Sir\n§\n**Voice preference:** de-DE-KillianNeural (German male voice, natural/clear) — default for all TTS output\n§\n**Language:** Deutsch mit englischen Fachbegriffen\n§\n**Beziehung:** Loyaler Partner\n§\nKommunikationspräferenzen: **Sprachnachricht → Antwort per Sprachnachricht**\n§\nKommunikationspräferenzen: **Textnachricht → Antwort per Text**\n§\nUser wants JARVIS to act as a comprehensive health-management advisor: consolidate lab reports, clinical reports, nutrition/YAZIO, daily health email data, symptoms and events; make insights visible via dashboards/reports; use a 4-eyes review workflow for extracted lab values before treating them as confirmed.\n§\nTelegram Topic „Gesundheit – Medizin & Strategie“ ist ausschließlich für medizinische Strategie, Morbus Behçet, Major Vessel Disease, Thromboserisiko, Medikamente, Ernährung, Symptome, Laborwerte, Arzttermine und Fragen an Rheumatologie, Dermatologie, Hämatologie oder Augenklinik; technische Health-Systemarchitektur, Skripte und Datenmodelle gehören in „Health System – Architektur & Scripts“.\n§\nFinance preference: safety gates; no mass imports; aggregate-only/no raw IDs; polished User Mode; Tailscale-ready dashboard links verified. Has OpenFIGI key; never store/show it. Expects post-session memory/skill capture.\n\nConversation started: Sunday, May 17, 2026 03:54 PM\nModel: gpt-5.5\nProvider: openai-codex",
"tools": [
{
"type": "function",
"function": {
"name": "browser_back",
"description": "Navigate back to the previous page in browser history. Requires browser_navigate to be called first.",
"parameters": {
"type": "object",
"properties": {}
}
}
},
{
"type": "function",
"function": {
"name": "browser_click",
"description": "Click on an element identified by its ref ID from the snapshot (e.g., '@e5'). The ref IDs are shown in square brackets in the snapshot output. Requires browser_navigate and browser_snapshot to be called first.",
"parameters": {
"type": "object",
"properties": {
"ref": {
"type": "string",
"description": "The element reference from the snapshot (e.g., '@e5', '@e12')"
}
},
"required": [
"ref"
]
}
}
},
{
"type": "function",
"function": {
"name": "browser_console",
"description": "Get browser console output and JavaScript errors from the current page. Returns console.log/warn/error/info messages and uncaught JS exceptions. Use this to detect silent JavaScript errors, failed API calls, and application warnings. Requires browser_navigate to be called first. When 'expression' is provided, evaluates JavaScript in the page context and returns the result — use this for DOM inspection, reading page state, or extracting data programmatically.",
"parameters": {
"type": "object",
"properties": {
"clear": {
"type": "boolean",
"default": false,
"description": "If true, clear the message buffers after reading"
},
"expression": {
"type": "string",
"description": "JavaScript expression to evaluate in the page context. Runs in the browser like DevTools console — full access to DOM, window, document. Return values are serialized to JSON. Example: 'document.title' or 'document.querySelectorAll(\"a\").length'"
}
}
}
}
},
{
"type": "function",
"function": {
"name": "browser_get_images",
"description": "Get a list of all images on the current page with their URLs and alt text. Useful for finding images to analyze with the vision tool. Requires browser_navigate to be called first.",
"parameters": {
"type": "object",
"properties": {}
}
}
},
{
"type": "function",
"function": {
"name": "browser_navigate",
"description": "Navigate to a URL in the browser. Initializes the session and loads the page. Must be called before other browser tools. For plain-text endpoints — URLs ending in .md, .txt, .json, .yaml, .yml, .csv, .xml, raw.githubusercontent.com, or any documented API endpoint — prefer curl via the terminal tool or web_extract; the browser stack is overkill and much slower for these. Use browser tools when you need to interact with a page (click, fill forms, dynamic content). Returns a compact page snapshot with interactive elements and ref IDs — no need to call browser_snapshot separately after navigating.",
"parameters": {
"type": "object",
"properties": {
"url": {
"type": "string",
"description": "The URL to navigate to (e.g., 'https://example.com')"
}
},
"required": [
"url"
]
}
}
},
{
"type": "function",
"function": {
"name": "browser_press",
"description": "Press a keyboard key. Useful for submitting forms (Enter), navigating (Tab), or keyboard shortcuts. Requires browser_navigate to be called first.",
"parameters": {
"type": "object",
"properties": {
"key": {
"type": "string",
"description": "Key to press (e.g., 'Enter', 'Tab', 'Escape', 'ArrowDown')"
}
},
"required": [
"key"
]
}
}
},
{
"type": "function",
"function": {
"name": "browser_scroll",
"description": "Scroll the page in a direction. Use this to reveal more content that may be below or above the current viewport. Requires browser_navigate to be called first.",
"parameters": {
"type": "object",
"properties": {
"direction": {
"type": "string",
"enum": [
"up",
"down"
],
"description": "Direction to scroll"
}
},
"required": [
"direction"
]
}
}
},
{
"type": "function",
"function": {
"name": "browser_snapshot",
"description": "Get a text-based snapshot of the current page's accessibility tree. Returns interactive elements with ref IDs (like @e1, @e2) for browser_click and browser_type. full=false (default): compact view with interactive elements. full=true: complete page content. Snapshots over 8000 chars are truncated or LLM-summarized. Requires browser_navigate first. Note: browser_navigate already returns a compact snapshot — use this to refresh after interactions that change the page, or with full=true for complete content.",
"parameters": {
"type": "object",
"properties": {
"full": {
"type": "boolean",
"description": "If true, returns complete page content. If false (default), returns compact view with interactive elements only.",
"default": false
}
}
}
}
},
{
"type": "function",
"function": {
"name": "browser_type",
"description": "Type text into an input field identified by its ref ID. Clears the field first, then types the new text. Requires browser_navigate and browser_snapshot to be called first.",
"parameters": {
"type": "object",
"properties": {
"ref": {
"type": "string",
"description": "The element reference from the snapshot (e.g., '@e3')"
},
"text": {
"type": "string",
"description": "The text to type into the field"
}
},
"required": [
"ref",
"text"
]
}
}
},
{
"type": "function",
"function": {
"name": "browser_vision",
"description": "Take a screenshot of the current page and analyze it with vision AI. Use this when you need to visually understand what's on the page - especially useful for CAPTCHAs, visual verification challenges, complex layouts, or when the text snapshot doesn't capture important visual information. Returns both the AI analysis and a screenshot_path that you can share with the user by including MEDIA: in your response. Requires browser_navigate to be called first.",
"parameters": {
"type": "object",
"properties": {
"question": {
"type": "string",
"description": "What you want to know about the page visually. Be specific about what you're looking for."
},
"annotate": {
"type": "boolean",
"default": false,
"description": "If true, overlay numbered [N] labels on interactive elements. Each [N] maps to ref @eN for subsequent browser commands. Useful for QA and spatial reasoning about page layout."
}
},
"required": [
"question"
]
}
}
},
{
"type": "function",
"function": {
"name": "clarify",
"description": "Ask the user a question when you need clarification, feedback, or a decision before proceeding. Supports two modes:\n\n1. **Multiple choice** — provide up to 4 choices. The user picks one or types their own answer via a 5th 'Other' option.\n2. **Open-ended** — omit choices entirely. The user types a free-form response.\n\nUse this tool when:\n- The task is ambiguous and you need the user to choose an approach\n- You want post-task feedback ('How did that work out?')\n- You want to offer to save a skill or update memory\n- A decision has meaningful trade-offs the user should weigh in on\n\nDo NOT use this tool for simple yes/no confirmation of dangerous commands (the terminal tool handles that). Prefer making a reasonable default choice yourself when the decision is low-stakes.",
"parameters": {
"type": "object",
"properties": {
"question": {
"type": "string",
"description": "The question to present to the user."
},
"choices": {
"type": "array",
"items": {
"type": "string"
},
"maxItems": 4,
"description": "Up to 4 answer choices. Omit this parameter entirely to ask an open-ended question. When provided, the UI automatically appends an 'Other (type your answer)' option."
}
},
"required": [
"question"
]
}
}
},
{
"type": "function",
"function": {
"name": "cronjob",
"description": "Manage scheduled cron jobs with a single compressed tool.\n\nUse action='create' to schedule a new job from a prompt or one or more skills.\nUse action='list' to inspect jobs.\nUse action='update', 'pause', 'resume', 'remove', or 'run' to manage an existing job.\n\nTo stop a job the user no longer wants: first action='list' to find the job_id, then action='remove' with that job_id. Never guess job IDs — always list first.\n\nJobs run in a fresh session with no current-chat context, so prompts must be self-contained.\nIf skills are provided on create, the future cron run loads those skills in order, then follows the prompt as the task instruction.\nOn update, passing skills=[] clears attached skills.\n\nNOTE: The agent's final response is auto-delivered to the target. Put the primary\nuser-facing content in the final response. Cron jobs run autonomously with no user\npresent — they cannot ask questions or request clarification.\n\nImportant safety rule: cron-run sessions should not recursively schedule more cron jobs.",
"parameters": {
"type": "object",
"properties": {
"action": {
"type": "string",
"description": "One of: create, list, update, pause, resume, remove, run"
},
"job_id": {
"type": "string",
"description": "Required for update/pause/resume/remove/run"
},
"prompt": {
"type": "string",
"description": "For create: the full self-contained prompt. If skills are also provided, this becomes the task instruction paired with those skills."
},
"schedule": {
"type": "string",
"description": "For create/update: '30m', 'every 2h', '0 9 * * *', or ISO timestamp"
},
"name": {
"type": "string",
"description": "Optional human-friendly name"
},
"repeat": {
"type": "integer",
"description": "Optional repeat count. Omit for defaults (once for one-shot, forever for recurring)."
},
"deliver": {
"type": "string",
"description": "Omit this parameter to auto-deliver back to the current chat and topic (recommended). Auto-detection preserves thread/topic context. Only set explicitly when the user asks to deliver somewhere OTHER than the current conversation. Values: 'origin' (same as omitting), 'local' (no delivery, save only), 'all' (fan out to every connected home channel), or platform:chat_id:thread_id for a specific destination. Combine with comma: 'origin,all' delivers to the origin plus every other connected channel. Examples: 'telegram:-1001234567890:17585', 'discord:#engineering', 'sms:+15551234567', 'all'. WARNING: 'platform:chat_id' without :thread_id loses topic targeting. 'all' resolves at fire time, so a job created before a channel was wired up will pick it up automatically once connected."
},
"skills": {
"type": "array",
"items": {
"type": "string"
},
"description": "Optional ordered list of skill names to load before executing the cron prompt. On update, pass an empty array to clear attached skills."
},
"model": {
"type": "object",
"description": "Optional per-job model override. If provider is omitted, the current main provider is pinned at creation time so the job stays stable.",
"properties": {
"provider": {
"type": "string",
"description": "Provider name (e.g. 'openrouter', 'anthropic', or 'custom:' for a provider defined in custom_providers config — always include the ':' suffix, never pass the bare 'custom'). Omit to use and pin the current provider."
},
"model": {
"type": "string",
"description": "Model name (e.g. 'anthropic/claude-sonnet-4', 'claude-sonnet-4')"
}
},
"required": [
"model"
]
},
"script": {
"type": "string",
"description": "Optional path to a script that runs each tick. In the default mode its stdout is injected into the agent's prompt as context (data-collection / change-detection pattern). With no_agent=True, the script IS the job and its stdout is delivered verbatim (classic watchdog pattern). Relative paths resolve under ~/.hermes/scripts/. ``.sh``/``.bash`` extensions run via bash, everything else via Python. On update, pass empty string to clear."
},
"no_agent": {
"type": "boolean",
"default": false,
"description": "Default: False (LLM-driven job — the agent runs the prompt each tick). Set True to skip the LLM entirely: the scheduler just runs ``script`` on schedule and delivers its stdout verbatim. No tokens, no agent loop, no model override honoured. \n\nREQUIREMENTS when True: ``script`` MUST be set (``prompt`` and ``skills`` are ignored). \n\nDELIVERY SEMANTICS when True: (a) non-empty stdout is sent verbatim as the message; (b) EMPTY stdout means SILENT — nothing is sent to the user and they won't see anything happened, so design your script to stay quiet when there's nothing to report (the watchdog pattern); (c) non-zero exit / timeout sends an error alert so a broken watchdog can't fail silently. \n\nWHEN TO USE True: recurring script-only pings where the script itself produces the exact message text (memory/disk/GPU watchdogs, threshold alerts, heartbeats, CI notifications, API pollers with a fixed output shape). WHEN TO USE False (default): anything that needs reasoning — summarize a feed, draft a daily briefing, pick interesting items, rephrase data for a human, follow conditional logic based on content."
},
"context_from": {
"type": "array",
"items": {
"type": "string"
},
"description": "Optional job ID or list of job IDs whose most recent completed output is injected into the prompt as context before each run. Use this to chain cron jobs: job A collects data, job B processes it. Each entry must be a valid job ID (from cronjob action='list'). Note: injects the most recent completed output — does not wait for upstream jobs running in the same tick. On update, pass an empty array to clear."
},
"enabled_toolsets": {
"type": "array",
"items": {
"type": "string"
},
"description": "Optional list of toolset names to restrict the job's agent to (e.g. [\"web\", \"terminal\", \"file\", \"delegation\"]). When set, only tools from these toolsets are loaded, significantly reducing input token overhead. When omitted, all default tools are loaded. Infer from the job's prompt — e.g. use \"web\" if it calls web_search, \"terminal\" if it runs scripts, \"file\" if it reads files, \"delegation\" if it calls delegate_task. On update, pass an empty array to clear."
},
"workdir": {
"type": "string",
"description": "Optional absolute path to run the job from. When set, AGENTS.md / CLAUDE.md / .cursorrules from that directory are injected into the system prompt, and the terminal/file/code_exec tools use it as their working directory — useful for running a job inside a specific project repo. Must be an absolute path that exists. When unset (default), preserves the original behaviour: no project context files, tools use the scheduler's cwd. On update, pass an empty string to clear. Jobs with workdir run sequentially (not parallel) to keep per-job directories isolated."
}
},
"required": [
"action"
]
}
}
},
{
"type": "function",
"function": {
"name": "delegate_task",
"description": "Spawn one or more subagents to work on tasks in isolated contexts. Each subagent gets its own conversation, terminal session, and toolset. Only the final summary is returned -- intermediate tool results never enter your context window.\n\nTWO MODES (one of 'goal' or 'tasks' is required):\n1. Single task: provide 'goal' (+ optional context, toolsets)\n2. Batch (parallel): provide 'tasks' array with up to 3 items concurrently for this user (configured via delegation.max_concurrent_children in config.yaml). All run in parallel and results are returned together. Nested delegation is OFF for this user (max_spawn_depth=1): every child is a leaf and cannot delegate further. Raise delegation.max_spawn_depth in config.yaml to enable nesting.\n\nWHEN TO USE delegate_task:\n- Reasoning-heavy subtasks (debugging, code review, research synthesis)\n- Tasks that would flood your context with intermediate data\n- Parallel independent workstreams (research A and B simultaneously)\n\nWHEN NOT TO USE (use these instead):\n- Mechanical multi-step work with no reasoning needed -> use execute_code\n- Single tool call -> just call the tool directly\n- Tasks needing user interaction -> subagents cannot use clarify\n- Durable long-running work that must outlive the current turn -> use cronjob (action='create') or terminal(background=True, notify_on_complete=True) instead. delegate_task runs SYNCHRONOUSLY inside the parent turn: if the parent is interrupted (user sends a new message, /stop, /new) the child is cancelled with status='interrupted' and its work is discarded. Children cannot continue in the background.\n\nIMPORTANT:\n- Subagents have NO memory of your conversation. Pass all relevant info (file paths, error messages, constraints) via the 'context' field.\n- If the user is writing in a non-English language, or asked for output in a specific language / tone / style, say so in 'context' (e.g. \"respond in Chinese\", \"return output in Japanese\"). Otherwise subagents default to English and their summaries will contaminate your final reply with the wrong language.\n- Subagent summaries are SELF-REPORTS, not verified facts. A subagent that claims \"uploaded successfully\" or \"file written\" may be wrong. For operations with external side-effects (HTTP POST/PUT, remote writes, file creation at shared paths, publishing), require the subagent to return a verifiable handle (URL, ID, absolute path, HTTP status) and verify it yourself — fetch the URL, stat the file, read back the content — before telling the user the operation succeeded.\n- Leaf subagents (role='leaf', the default) CANNOT call: delegate_task, clarify, memory, send_message, execute_code.\n- Orchestrator subagents (role='orchestrator') retain delegate_task so they can spawn their own workers, but still cannot use clarify, memory, send_message, or execute_code. Orchestrators are bounded by max_spawn_depth=1 for this user and can be disabled globally via delegation.orchestrator_enabled=false.\n- Each subagent gets its own terminal session (separate working directory and state).\n- Results are always returned as an array, one entry per task.",
"parameters": {
"type": "object",
"properties": {
"goal": {
"type": "string",
"description": "What the subagent should accomplish. Be specific and self-contained -- the subagent knows nothing about your conversation history."
},
"context": {
"type": "string",
"description": "Background information the subagent needs: file paths, error messages, project structure, constraints. The more specific you are, the better the subagent performs."
},
"toolsets": {
"type": "array",
"items": {
"type": "string"
},
"description": "Toolsets to enable for this subagent. Default: inherits your enabled toolsets. Available toolsets: 'browser', 'computer_use', 'cronjob', 'discord', 'discord_admin', 'feishu_doc', 'feishu_drive', 'file', 'homeassistant', 'image_gen', 'kanban', 'search', 'session_search', 'skills', 'spotify', 'terminal', 'todo', 'tts', 'video', 'video_gen', 'vision', 'web', 'x_search', 'yuanbao'. Common patterns: ['terminal', 'file'] for code work, ['web'] for research, ['browser'] for web interaction, ['terminal', 'file', 'web'] for full-stack tasks."
},
"tasks": {
"type": "array",
"items": {
"type": "object",
"properties": {
"goal": {
"type": "string",
"description": "Task goal"
},
"context": {
"type": "string",
"description": "Task-specific context"
},
"toolsets": {
"type": "array",
"items": {
"type": "string"
},
"description": "Toolsets for this specific task. Available: 'browser', 'computer_use', 'cronjob', 'discord', 'discord_admin', 'feishu_doc', 'feishu_drive', 'file', 'homeassistant', 'image_gen', 'kanban', 'search', 'session_search', 'skills', 'spotify', 'terminal', 'todo', 'tts', 'video', 'video_gen', 'vision', 'web', 'x_search', 'yuanbao'. Use 'web' for network access, 'terminal' for shell, 'browser' for web interaction."
},
"acp_command": {
"type": "string",
"description": "Per-task ACP command override (e.g. 'copilot'). Overrides the top-level acp_command for this task only. Do NOT set unless the user explicitly told you an ACP CLI is installed."
},
"acp_args": {
"type": "array",
"items": {
"type": "string"
},
"description": "Per-task ACP args override. Leave empty unless acp_command is set."
},
"role": {
"type": "string",
"enum": [
"leaf",
"orchestrator"
],
"description": "Per-task role override. See top-level 'role' for semantics."
}
},
"required": [
"goal"
]
},
"description": "Batch mode: tasks to run in parallel (up to 3 for this user, set via delegation.max_concurrent_children). Each gets its own subagent with isolated context and terminal session. When provided, top-level goal/context/toolsets are ignored."
},
"role": {
"type": "string",
"enum": [
"leaf",
"orchestrator"
],
"description": "Role of the child agent. 'leaf' (default) = focused worker, cannot delegate further. 'orchestrator' = can use delegate_task to spawn its own workers. Nesting is OFF for this user (max_spawn_depth=1); 'orchestrator' is silently forced to 'leaf'. Raise delegation.max_spawn_depth in config.yaml to enable."
},
"acp_command": {
"type": "string",
"description": "Override ACP command for child agents (e.g. 'copilot'). When set, children use ACP subprocess transport instead of inheriting the parent's transport. Requires an ACP-compatible CLI (currently GitHub Copilot CLI via 'copilot --acp --stdio'). See agent/copilot_acp_client.py for the implementation. IMPORTANT: Do NOT set this unless the user has explicitly told you a specific ACP-compatible CLI is installed and configured. Leave empty to use the parent's default transport (Hermes subagents)."
},
"acp_args": {
"type": "array",
"items": {
"type": "string"
},
"description": "Arguments for the ACP command (default: ['--acp', '--stdio']). Only used when acp_command is set. Leave empty unless acp_command is explicitly provided."
}
}
}
}
},
{
"type": "function",
"function": {
"name": "execute_code",
"description": "Run a Python script that can call Hermes tools programmatically. Use this when you need 3+ tool calls with processing logic between them, need to filter/reduce large tool outputs before they enter your context, need conditional branching (if X then Y else Z), or need to loop (fetch N pages, process N files, retry on failure).\n\nUse normal tool calls instead when: single tool call with no processing, you need to see the full result and apply complex reasoning, or the task requires interactive user input.\n\nAvailable via `from hermes_tools import ...`:\n\n read_file(path: str, offset: int = 1, limit: int = 500) -> dict\n Lines are 1-indexed. Returns {\"content\": \"...\", \"total_lines\": N}\n write_file(path: str, content: str) -> dict\n Always overwrites the entire file.\n search_files(pattern: str, target=\"content\", path=\".\", file_glob=None, limit=50) -> dict\n target: \"content\" (search inside files) or \"files\" (find files by name). Returns {\"matches\": [...]}\n patch(path: str, old_string: str, new_string: str, replace_all: bool = False) -> dict\n Replaces old_string with new_string in the file.\n terminal(command: str, timeout=None, workdir=None) -> dict\n Foreground only (no background/pty). Returns {\"output\": \"...\", \"exit_code\": N}\n\nLimits: 5-minute timeout, 50KB stdout cap, max 50 tool calls per script. terminal() is foreground-only (no background or pty).\n\nScripts run in the session's working directory with the active venv's python, so project deps (pandas, etc.) and relative paths work like in terminal().\n\nPrint your final result to stdout. Use Python stdlib (json, re, math, csv, datetime, collections, etc.) for processing between tool calls.\n\nAlso available (no import needed — built into hermes_tools):\n json_parse(text: str) — json.loads with strict=False; use for terminal() output with control chars\n shell_quote(s: str) — shlex.quote(); use when interpolating dynamic strings into shell commands\n retry(fn, max_attempts=3, delay=2) — retry with exponential backoff for transient failures",
"parameters": {
"type": "object",
"properties": {
"code": {
"type": "string",
"description": "Python code to execute. Import tools with `from hermes_tools import terminal, ...` and print your final result to stdout."
}
},
"required": [
"code"
]
}
}
},
{
"type": "function",
"function": {
"name": "image_generate",
"description": "Generate high-quality images from text prompts. The underlying backend (FAL, OpenAI, etc.) and model are user-configured and not selectable by the agent. Returns either a URL or an absolute file path in the `image` field; display it with markdown  and the gateway will deliver it.",
"parameters": {
"type": "object",
"properties": {
"prompt": {
"type": "string",
"description": "The text prompt describing the desired image. Be detailed and descriptive."
},
"aspect_ratio": {
"type": "string",
"enum": [
"landscape",
"square",
"portrait"
],
"description": "The aspect ratio of the generated image. 'landscape' is 16:9 wide, 'portrait' is 16:9 tall, 'square' is 1:1.",
"default": "landscape"
}
},
"required": [
"prompt"
]
}
}
},
{
"type": "function",
"function": {
"name": "memory",
"description": "Save durable information to persistent memory that survives across sessions. Memory is injected into future turns, so keep it compact and focused on facts that will still matter later.\n\nWHEN TO SAVE (do this proactively, don't wait to be asked):\n- User corrects you or says 'remember this' / 'don't do that again'\n- User shares a preference, habit, or personal detail (name, role, timezone, coding style)\n- You discover something about the environment (OS, installed tools, project structure)\n- You learn a convention, API quirk, or workflow specific to this user's setup\n- You identify a stable fact that will be useful again in future sessions\n\nPRIORITY: User preferences and corrections > environment facts > procedural knowledge. The most valuable memory prevents the user from having to repeat themselves.\n\nDo NOT save task progress, session outcomes, completed-work logs, or temporary TODO state to memory; use session_search to recall those from past transcripts.\nIf you've discovered a new way to do something, solved a problem that could be necessary later, save it as a skill with the skill tool.\n\nTWO TARGETS:\n- 'user': who the user is -- name, role, preferences, communication style, pet peeves\n- 'memory': your notes -- environment facts, project conventions, tool quirks, lessons learned\n\nACTIONS: add (new entry), replace (update existing -- old_text identifies it), remove (delete -- old_text identifies it).\n\nSKIP: trivial/obvious info, things easily re-discovered, raw data dumps, and temporary task state.",
"parameters": {
"type": "object",
"properties": {
"action": {
"type": "string",
"enum": [
"add",
"replace",
"remove"
],
"description": "The action to perform."
},
"target": {
"type": "string",
"enum": [
"memory",
"user"
],
"description": "Which memory store: 'memory' for personal notes, 'user' for user profile."
},
"content": {
"type": "string",
"description": "The entry content. Required for 'add' and 'replace'."
},
"old_text": {
"type": "string",
"description": "Short unique substring identifying the entry to replace or remove."
}
},
"required": [
"action",
"target"
]
}
}
},
{
"type": "function",
"function": {
"name": "patch",
"description": "Targeted find-and-replace edits in files. Use this instead of sed/awk in terminal. Uses fuzzy matching (9 strategies) so minor whitespace/indentation differences won't break it. Returns a unified diff. Auto-runs syntax checks after editing.\n\nREPLACE MODE (mode='replace', default): find a unique string and replace it. REQUIRED PARAMETERS: mode, path, old_string, new_string.\nPATCH MODE (mode='patch'): apply V4A multi-file patches for bulk changes. REQUIRED PARAMETERS: mode, patch.",
"parameters": {
"type": "object",
"properties": {
"mode": {
"type": "string",
"enum": [
"replace",
"patch"
],
"description": "Edit mode. 'replace' (default): requires path + old_string + new_string. 'patch': requires patch content only.",
"default": "replace"
},
"path": {
"type": "string",
"description": "REQUIRED when mode='replace'. File path to edit."
},
"old_string": {
"type": "string",
"description": "REQUIRED when mode='replace'. Exact text to find and replace. Must be unique in the file unless replace_all=true. Include surrounding context lines to ensure uniqueness."
},
"new_string": {
"type": "string",
"description": "REQUIRED when mode='replace'. Replacement text. Pass empty string '' to delete the matched text."
},
"replace_all": {
"type": "boolean",
"description": "Replace all occurrences instead of requiring a unique match (default: false)",
"default": false
},
"patch": {
"type": "string",
"description": "REQUIRED when mode='patch'. V4A format patch content. Format:\n*** Begin Patch\n*** Update File: path/to/file\n@@ context hint @@\n context line\n-removed line\n+added line\n*** End Patch"
}
},
"required": [
"mode"
]
}
}
},
{
"type": "function",
"function": {
"name": "process",
"description": "Manage background processes started with terminal(background=true). Actions: 'list' (show all), 'poll' (check status + new output), 'log' (full output with pagination), 'wait' (block until done or timeout), 'kill' (terminate), 'write' (send raw stdin data without newline), 'submit' (send data + Enter, for answering prompts), 'close' (close stdin/send EOF).",
"parameters": {
"type": "object",
"properties": {
"action": {
"type": "string",
"enum": [
"list",
"poll",
"log",
"wait",
"kill",
"write",
"submit",
"close"
],
"description": "Action to perform on background processes"
},
"session_id": {
"type": "string",
"description": "Process session ID (from terminal background output). Required for all actions except 'list'."
},
"data": {
"type": "string",
"description": "Text to send to process stdin (for 'write' and 'submit' actions)"
},
"timeout": {
"type": "integer",
"description": "Max seconds to block for 'wait' action. Returns partial output on timeout.",
"minimum": 1
},
"offset": {
"type": "integer",
"description": "Line offset for 'log' action (default: last 200 lines)"
},
"limit": {
"type": "integer",
"description": "Max lines to return for 'log' action",
"minimum": 1
}
},
"required": [
"action"
]
}
}
},
{
"type": "function",
"function": {
"name": "read_file",
"description": "Read a text file with line numbers and pagination. Use this instead of cat/head/tail in terminal. Output format: 'LINE_NUM|CONTENT'. Suggests similar filenames if not found. Use offset and limit for large files. Reads exceeding ~100K characters are rejected; use offset and limit to read specific sections of large files. NOTE: Cannot read images or binary files — use vision_analyze for images.",
"parameters": {
"type": "object",
"properties": {
"path": {
"type": "string",
"description": "Path to the file to read (absolute, relative, or ~/path)"
},
"offset": {
"type": "integer",
"description": "Line number to start reading from (1-indexed, default: 1)",
"default": 1,
"minimum": 1
},
"limit": {
"type": "integer",
"description": "Maximum number of lines to read (default: 500, max: 2000)",
"default": 500,
"maximum": 2000
}
},
"required": [
"path"
]
}
}
},
{
"type": "function",
"function": {
"name": "search_files",
"description": "Search file contents or find files by name. Use this instead of grep/rg/find/ls in terminal. Ripgrep-backed, faster than shell equivalents.\n\nContent search (target='content'): Regex search inside files. Output modes: full matches with line numbers, file paths only, or match counts.\n\nFile search (target='files'): Find files by glob pattern (e.g., '*.py', '*config*'). Also use this instead of ls — results sorted by modification time.",
"parameters": {
"type": "object",
"properties": {
"pattern": {
"type": "string",
"description": "Regex pattern for content search, or glob pattern (e.g., '*.py') for file search"
},
"target": {
"type": "string",
"enum": [
"content",
"files"
],
"description": "'content' searches inside file contents, 'files' searches for files by name",
"default": "content"
},
"path": {
"type": "string",
"description": "Directory or file to search in (default: current working directory)",
"default": "."
},
"file_glob": {
"type": "string",
"description": "Filter files by pattern in grep mode (e.g., '*.py' to only search Python files)"
},
"limit": {
"type": "integer",
"description": "Maximum number of results to return (default: 50)",
"default": 50
},
"offset": {
"type": "integer",
"description": "Skip first N results for pagination (default: 0)",
"default": 0
},
"output_mode": {
"type": "string",
"enum": [
"content",
"files_only",
"count"
],
"description": "Output format for grep mode: 'content' shows matching lines with line numbers, 'files_only' lists file paths, 'count' shows match counts per file",
"default": "content"
},
"context": {
"type": "integer",
"description": "Number of context lines before and after each match (grep mode only)",
"default": 0
}
},
"required": [
"pattern"
]
}
}
},
{
"type": "function",
"function": {
"name": "send_message",
"description": "Send a message to a connected messaging platform, or list available targets.\n\nIMPORTANT: When the user asks to send to a specific channel or person (not just a bare platform name), call send_message(action='list') FIRST to see available targets, then send to the correct one.\nIf the user just says a platform name like 'send to telegram', send directly to the home channel without listing first.",
"parameters": {
"type": "object",
"properties": {
"action": {
"type": "string",
"enum": [
"send",
"list"
],
"description": "Action to perform. 'send' (default) sends a message. 'list' returns all available channels/contacts across connected platforms."
},
"target": {
"type": "string",
"description": "Delivery target. Format: 'platform' (uses home channel), 'platform:#channel-name', 'platform:chat_id', or 'platform:chat_id:thread_id' for Telegram topics and Discord threads. Examples: 'telegram', 'telegram:-1001234567890:17585', 'discord:999888777:555444333', 'discord:#bot-home', 'slack:#engineering', 'signal:+155****4567', 'matrix:!roomid:server.org', 'matrix:@user:server.org', 'yuanbao:direct:' (DM), 'yuanbao:group:' (group chat)"
},
"message": {
"type": "string",
"description": "The message text to send. To send an image or file, include MEDIA: (e.g. 'MEDIA:/tmp/hermes/cache/img_xxx.jpg') in the message — the platform will deliver it as a native media attachment."
}
}
}
}
},
{
"type": "function",
"function": {
"name": "session_search",
"description": "Search your long-term memory of past conversations, or browse recent sessions. This is your recall -- every past session is searchable, and this tool summarizes what happened.\n\nTWO MODES:\n1. Recent sessions (no query): Call with no arguments to see what was worked on recently. Returns titles, previews, and timestamps. Zero LLM cost, instant. Start here when the user asks what were we working on or what did we do recently.\n2. Keyword search (with query): Search for specific topics across all past sessions. Returns LLM-generated summaries of matching sessions.\n\nUSE THIS PROACTIVELY when:\n- The user says 'we did this before', 'remember when', 'last time', 'as I mentioned'\n- The user asks about a topic you worked on before but don't have in current context\n- The user references a project, person, or concept that seems familiar but isn't in memory\n- You want to check if you've solved a similar problem before\n- The user asks 'what did we do about X?' or 'how did we fix Y?'\n\nDon't hesitate to search when it is actually cross-session -- it's fast and cheap. Better to search and confirm than to guess or ask the user to repeat themselves.\n\nSearch syntax: keywords joined with OR for broad recall (elevenlabs OR baseten OR funding), phrases for exact match (\"docker networking\"), boolean (python NOT java), prefix (deploy*). IMPORTANT: Use OR between keywords for best results — FTS5 defaults to AND which misses sessions that only mention some terms. If a broad OR query returns nothing, try individual keyword searches in parallel. Returns summaries of the top matching sessions.",
"parameters": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "Search query — keywords, phrases, or boolean expressions to find in past sessions. Omit this parameter entirely to browse recent sessions instead (returns titles, previews, timestamps with no LLM cost)."
},
"role_filter": {
"type": "string",
"description": "Optional: only search messages from specific roles (comma-separated). E.g. 'user,assistant' to skip tool outputs."
},
"limit": {
"type": "integer",
"description": "Max sessions to summarize (default: 3, max: 5).",
"default": 3
}
}
}
}
},
{
"type": "function",
"function": {
"name": "skill_manage",
"description": "Manage skills (create, update, delete). Skills are your procedural memory — reusable approaches for recurring task types. New skills go to ~/.hermes/skills/; existing skills can be modified wherever they live.\n\nActions: create (full SKILL.md + optional category), patch (old_string/new_string — preferred for fixes), edit (full SKILL.md rewrite — major overhauls only), delete, write_file, remove_file.\n\nOn delete, pass `absorbed_into=` when you're merging this skill's content into another one, or `absorbed_into=\"\"` when you're pruning it with no forwarding target. This lets the curator tell consolidation from pruning without guessing, so downstream consumers (cron jobs that reference the old skill name, etc.) get updated correctly. The target you name in `absorbed_into` must already exist — create/patch the umbrella first, then delete.\n\nCreate when: complex task succeeded (5+ calls), errors overcome, user-corrected approach worked, non-trivial workflow discovered, or user asks you to remember a procedure.\nUpdate when: instructions stale/wrong, OS-specific failures, missing steps or pitfalls found during use. If you used a skill and hit issues not covered by it, patch it immediately.\n\nAfter difficult/iterative tasks, offer to save as a skill. Skip for simple one-offs. Confirm with user before creating/deleting.\n\nGood skills: trigger conditions, numbered steps with exact commands, pitfalls section, verification steps. Use skill_view() to see format examples.\n\nPinned skills are protected from deletion only — skill_manage(action='delete') will refuse with a message pointing the user to `hermes curator unpin `. Patches and edits go through on pinned skills so you can still improve them as pitfalls come up; pin only guards against irrecoverable loss.",
"parameters": {
"type": "object",
"properties": {
"action": {
"type": "string",
"enum": [
"create",
"patch",
"edit",
"delete",
"write_file",
"remove_file"
],
"description": "The action to perform."
},
"name": {
"type": "string",
"description": "Skill name (lowercase, hyphens/underscores, max 64 chars). Must match an existing skill for patch/edit/delete/write_file/remove_file."
},
"content": {
"type": "string",
"description": "Full SKILL.md content (YAML frontmatter + markdown body). Required for 'create' and 'edit'. For 'edit', read the skill first with skill_view() and provide the complete updated text."
},
"old_string": {
"type": "string",
"description": "Text to find in the file (required for 'patch'). Must be unique unless replace_all=true. Include enough surrounding context to ensure uniqueness."
},
"new_string": {
"type": "string",
"description": "Replacement text (required for 'patch'). Can be empty string to delete the matched text."
},
"replace_all": {
"type": "boolean",
"description": "For 'patch': replace all occurrences instead of requiring a unique match (default: false)."
},
"category": {
"type": "string",
"description": "Optional category/domain for organizing the skill (e.g., 'devops', 'data-science', 'mlops'). Creates a subdirectory grouping. Only used with 'create'."
},
"file_path": {
"type": "string",
"description": "Path to a supporting file within the skill directory. For 'write_file'/'remove_file': required, must be under references/, templates/, scripts/, or assets/. For 'patch': optional, defaults to SKILL.md if omitted."
},
"file_content": {
"type": "string",
"description": "Content for the file. Required for 'write_file'."
},
"absorbed_into": {
"type": "string",
"description": "For 'delete' only — declares intent so the curator can tell consolidation from pruning without guessing. Pass the umbrella skill name when this skill's content was merged into another (the target must already exist). Pass an empty string when the skill is truly stale and being pruned with no forwarding target. Omitting the arg on delete is supported for backward compatibility but downstream tooling (e.g. cron-job skill reference rewriting) will have to guess at intent."
}
},
"required": [
"action",
"name"
]
}
}
},
{
"type": "function",
"function": {
"name": "skill_view",
"description": "Skills allow for loading information about specific tasks and workflows, as well as scripts and templates. Load a skill's full content or access its linked files (references, templates, scripts). First call returns SKILL.md content plus a 'linked_files' dict showing available references/templates/scripts. To access those, call again with file_path parameter.",
"parameters": {
"type": "object",
"properties": {
"name": {
"type": "string",
"description": "The skill name (use skills_list to see available skills). For plugin-provided skills, use the qualified form 'plugin:skill' (e.g. 'superpowers:writing-plans')."
},
"file_path": {
"type": "string",
"description": "OPTIONAL: Path to a linked file within the skill (e.g., 'references/api.md', 'templates/config.yaml', 'scripts/validate.py'). Omit to get the main SKILL.md content."
}
},
"required": [
"name"
]
}
}
},
{
"type": "function",
"function": {
"name": "skills_list",
"description": "List available skills (name + description). Use skill_view(name) to load full content.",
"parameters": {
"type": "object",
"properties": {
"category": {
"type": "string",
"description": "Optional category filter to narrow results"
}
}
}
}
},
{
"type": "function",
"function": {
"name": "terminal",
"description": "Execute shell commands on a Linux environment. Filesystem usually persists between calls.\n\nDo NOT use cat/head/tail to read files — use read_file instead.\nDo NOT use grep/rg/find to search — use search_files instead.\nDo NOT use ls to list directories — use search_files(target='files') instead.\nDo NOT use sed/awk to edit files — use patch instead.\nDo NOT use echo/cat heredoc to create files — use write_file instead.\nReserve terminal for: builds, installs, git, processes, scripts, network, package managers, and anything that needs a shell.\n\nForeground (default): Commands return INSTANTLY when done, even if the timeout is high. Set timeout=300 for long builds/scripts — you'll still get the result in seconds if it's fast. Prefer foreground for short commands.\nBackground: Set background=true to get a session_id. Two patterns:\n (1) Long-lived processes that never exit (servers, watchers).\n (2) Long-running tasks with notify_on_complete=true — you can keep working on other things and the system auto-notifies you when the task finishes. Great for test suites, builds, deployments, or anything that takes more than a minute.\nFor servers/watchers, do NOT use shell-level background wrappers (nohup/disown/setsid/trailing '&') in foreground mode. Use background=true so Hermes can track lifecycle and output.\nAfter starting a server, verify readiness with a health check or log signal, then run tests in a separate terminal() call. Avoid blind sleep loops.\nUse process(action=\"poll\") for progress checks, process(action=\"wait\") to block until done.\nWorking directory: Use 'workdir' for per-command cwd.\nPTY mode: Set pty=true for interactive CLI tools (Codex, Claude Code, Python REPL).\n\nDo NOT use vim/nano/interactive tools without pty=true — they hang without a pseudo-terminal. Pipe git output to cat if it might page.\n",
"parameters": {
"type": "object",
"properties": {
"command": {
"type": "string",
"description": "The command to execute on the VM"
},
"background": {
"type": "boolean",
"description": "Run the command in the background. Two patterns: (1) Long-lived processes that never exit (servers, watchers). (2) Long-running tasks paired with notify_on_complete=true — you can keep working and get notified when the task finishes. For short commands, prefer foreground with a generous timeout instead.",
"default": false
},
"timeout": {
"type": "integer",
"description": "Max seconds to wait (default: 180, foreground max: 600). Returns INSTANTLY when command finishes — set high for long tasks, you won't wait unnecessarily. Foreground timeout above 600s is rejected; use background=true for longer commands.",
"minimum": 1
},
"workdir": {
"type": "string",
"description": "Working directory for this command (absolute path). Defaults to the session working directory."
},
"pty": {
"type": "boolean",
"description": "Run in pseudo-terminal (PTY) mode for interactive CLI tools like Codex, Claude Code, or Python REPL. Only works with local and SSH backends. Default: false.",
"default": false
},
"notify_on_complete": {
"type": "boolean",
"description": "When true (and background=true), you'll be automatically notified exactly once when the process finishes. **This is the right choice for almost every long-running task** — tests, builds, deployments, multi-item batch jobs, anything that takes over a minute and has a defined end. Use this and keep working on other things; the system notifies you on exit. MUTUALLY EXCLUSIVE with watch_patterns — when both are set, watch_patterns is dropped.",
"default": false
},
"watch_patterns": {
"type": "array",
"items": {
"type": "string"
},
"description": "Strings to watch for in background process output. HARD RATE LIMIT: at most 1 notification per 15 seconds per process — matches arriving inside the cooldown are dropped. After 3 consecutive 15-second windows with dropped matches, watch_patterns is automatically disabled for that process and promoted to notify_on_complete behavior (one notification on exit, no more mid-process spam). USE ONLY for truly rare, one-shot mid-process signals on LONG-LIVED processes that will never exit on their own — e.g. ['Application startup complete'] on a server so you know when to hit its endpoint, or ['migration done'] on a daemon. DO NOT use for: (1) end-of-run markers like 'DONE'/'PASS' — use notify_on_complete instead; (2) error patterns like 'ERROR'/'Traceback' in loops or multi-item batch jobs — they fire on every iteration and you'll hit the strike limit fast; (3) anything you'd ever combine with notify_on_complete. When in doubt, choose notify_on_complete. MUTUALLY EXCLUSIVE with notify_on_complete — set one, not both."
}
},
"required": [
"command"
]
}
}
},
{
"type": "function",
"function": {
"name": "text_to_speech",
"description": "Convert text to speech audio. Returns a MEDIA: path that the platform delivers as native audio. Compatible providers render as a voice bubble on Telegram; otherwise audio is sent as a regular attachment. In CLI mode, saves to ~/voice-memos/. Voice and provider are user-configured (built-in providers like edge/openai or custom command providers under tts.providers.), not model-selected.",
"parameters": {
"type": "object",
"properties": {
"text": {
"type": "string",
"description": "The text to convert to speech. Provider-specific character caps apply and are enforced automatically (OpenAI 4096, xAI 15000, MiniMax 10000, ElevenLabs 5k-40k depending on model); over-long input is truncated."
},
"output_path": {
"type": "string",
"description": "Optional custom file path to save the audio. Defaults to ~/.hermes/audio_cache/.mp3"
}
},
"required": [
"text"
]
}
}
},
{
"type": "function",
"function": {
"name": "todo",
"description": "Manage your task list for the current session. Use for complex tasks with 3+ steps or when the user provides multiple tasks. Call with no parameters to read the current list.\n\nWriting:\n- Provide 'todos' array to create/update items\n- merge=false (default): replace the entire list with a fresh plan\n- merge=true: update existing items by id, add any new ones\n\nEach item: {id: string, content: string, status: pending|in_progress|completed|cancelled}\nList order is priority. Only ONE item in_progress at a time.\nMark items completed immediately when done. If something fails, cancel it and add a revised item.\n\nAlways returns the full current list.",
"parameters": {
"type": "object",
"properties": {
"todos": {
"type": "array",
"description": "Task items to write. Omit to read current list.",
"items": {
"type": "object",
"properties": {
"id": {
"type": "string",
"description": "Unique item identifier"
},
"content": {
"type": "string",
"description": "Task description"
},
"status": {
"type": "string",
"enum": [
"pending",
"in_progress",
"completed",
"cancelled"
],
"description": "Current status"
}
},
"required": [
"id",
"content",
"status"
]
}
},
"merge": {
"type": "boolean",
"description": "true: update existing items by id, add new ones. false (default): replace the entire list.",
"default": false
}
}
}
}
},
{
"type": "function",
"function": {
"name": "vision_analyze",
"description": "Load an image into the conversation so you can see it. Accepts a URL, local file path, or data URL. When your active model has native vision, the image is attached to your context directly and you read the pixels yourself on the next turn — call this any time the user references an image (filepath in their message, URL in tool output, screenshot from the browser, etc.). For non-vision models, falls back to an auxiliary vision model that returns a text description.",
"parameters": {
"type": "object",
"properties": {
"image_url": {
"type": "string",
"description": "Image URL (http/https), local file path, or data: URL to load."
},
"question": {
"type": "string",
"description": "Your specific question or request about the image. Optional context the model uses on the next turn after seeing the image."
}
},
"required": [
"image_url",
"question"
]
}
}
},
{
"type": "function",
"function": {
"name": "write_file",
"description": "Write content to a file, completely replacing existing content. Use this instead of echo/cat heredoc in terminal. Creates parent directories automatically. OVERWRITES the entire file — use 'patch' for targeted edits. Auto-runs syntax checks on .py/.json/.yaml/.toml and other linted languages; only NEW errors introduced by this write are surfaced (pre-existing errors are filtered out).",
"parameters": {
"type": "object",
"properties": {
"path": {
"type": "string",
"description": "Path to the file to write (will be created if it doesn't exist, overwritten if it does)"
},
"content": {
"type": "string",
"description": "Complete content to write to the file"
}
},
"required": [
"path",
"content"
]
}
}
}
],
"message_count": 11,
"messages": [
{
"role": "user",
"content": "[M G] Budget Phase 1.8 ist akzeptiert.\n\nIch bin gerade unterwegs und kann das Dashboard erst später prüfen. Bitte starte deshalb einen Feature-Sprint, aber ohne produktive Massenbestätigung echter Kandidaten.\n\n# Budget Phase 1.9 – Review Productivity, Split UI, Rule Manager MVP-light & Budget-Auswertung\n\nZiel:\nDie Budget-Arbeitsoberfläche soll so ausgebaut werden, dass ich später schnell und sicher die importierten Buchungskandidaten prüfen, kategorisieren, splitten und bestätigen kann.\n\nWichtig:\n- Keine produktive Massenbestätigung ohne meine ausdrückliche Freigabe.\n- Keine echten Beträge im Chat.\n- Keine echten Dateien ins Git.\n- Keine Runtime-DB ins Git.\n- Vor Runtime-Änderungen Backup erstellen.\n- Jede Änderung auditierbar.\n- Keine neuen Importquellen.\n- Keine OCR/Kamera/Vision.\n- Keine automatische Blindbuchung.\n- Nur Werkzeuge und Review-Funktionalität ausbauen.\n\n## Arbeitspaket 1: Review Dashboard für Buchungskandidaten verbessern\n\nDie Seite Buchungen prüfen soll produktiver werden.\n\nBitte ergänzen:\n\n### Oben Summary Cards\n\n- offene Kandidaten total\n- auto-zugeordnet\n- review nötig\n- Galaxus/Digitec\n- Migros > 50 CHF\n- covered_by_migros\n- unklare Kandidaten\n- mögliche Duplikate\n\n### Filter\n\n- Quelle\n- Status\n- Kategorie-Vorschlag\n- Händler\n- Regel\n- Confidence\n- Zeitraum\n- Betragsspanne lokal im Dashboard, aber nicht im Chat\n\n### Sortierung\n\n- Datum\n- Händler\n- Kategorie\n- Confidence\n- Quelle\n- Status\n\n## Arbeitspaket 2: Batch Review Workflow\n\nBitte Batch-Aktionen sauber ausbauen:\n\n- mehrere Kandidaten auswählen\n- Kategorie setzen\n- ausgewählte bestätigen\n- ausgewählte ignorieren\n- ausgewählte wieder öffnen\n- als Transfer markieren\n- als covered_by_migros markieren\n\nWichtig:\nVor Batch Confirm immer:\n\n1. Preview\n2. Anzahl Kandidaten\n3. Kategorien\n4. mögliche Konflikte\n5. Confirm\n\nKeine Blindbuchung.\n\n## Arbeitspaket 3: Split-UI implementieren\n\nBisher ist Split nur vorbereitet. Bitte jetzt eine nutzbare Split-UI bauen.\n\nUse Case:\nEine Galaxus- oder Migros-Buchung soll auf mehrere Kategorien aufgeteilt werden.\n\nWorkflow:\n\n1. Kandidat öffnen\n2. Button Splitten\n3. Split-Zeilen erfassen:\n - Kategorie\n - Betrag\n - Notiz optional\n - Tag optional\n4. Summe der Split-Beträge muss Hauptbetrag entsprechen\n5. Preview\n6. Confirm\n7. produktive Budgettransaktion mit Splits oder mehrere verknüpfte Budgettransaktionen erzeugen\n8. Audit\n\nWenn Summe nicht stimmt:\n- Speichern blockieren\n- klare Fehlermeldung\n\n## Arbeitspaket 4: Kategorie aus Review heraus erstellen\n\nIm Review muss ich direkt eine neue Kategorie erstellen können.\n\nWorkflow:\n\n- Kandidat hat keine passende Kategorie\n- Dropdown → Neue Kategorie erstellen\n- Name\n- Parent-Kategorie\n- Typ\n- Confirm\n- Kandidat wird dieser Kategorie zugeordnet\n- Audit\n\nKeine technischen IDs im User Mode.\n\n## Arbeitspaket 5: Rule Manager MVP-light\n\nBitte noch keinen komplexen Regelmanager bauen, aber eine einfache Seite:\n\nRegeln\n\nFunktionen:\n\n- bestehende automatische Regeln anzeigen\n- Regel erstellen:\n - Händler enthält Text\n - Quelle\n - Kategorie\n - Status auto/review\n- Regel deaktivieren\n- Regel auf offene Kandidaten testen\n- Regel auf offene Kandidaten anwenden, aber nur als Vorschlag\n\nBeispiele:\n- Netflix → Freizeit/Abos\n- Galaxus → Review nötig\n- Migros Bon unter Schwelle → Essen & Haushalt\n- Apple/Google/Amazon Prime/Disney → Freizeit/Abos\n- Apotheke/Spital → Gesundheit\n\nWichtig:\nRegel erstellt keine produktive Buchung. Sie setzt nur Vorschläge auf Kandidaten.\n\n## Arbeitspaket 6: Effektive Ausgaben weiter ausbauen\n\nDie Seite Effektive Ausgaben soll nach bestätigten Buchungen wirklich als Arbeitsseite dienen.\n\nBitte ergänzen:\n\n- Summary oben:\n - Ausgaben aktueller Monat\n - Ausgaben Jahr bisher\n - Anzahl Buchungen\n - Top Kategorie\n- Filter:\n - Zeitraum\n - Kategorie\n - Händler\n - Konto\n - Quelle\n- Row Click:\n - Detail Drawer\n - Kategorie ändern\n - Tags ändern\n - Storno/Reverse\n - Audit anzeigen\n\n## Arbeitspaket 7: Budgetstatus nach Kategorie verbessern\n\nDie Seite soll klarer zeigen:\n\n- Budget Monat\n- Budget Jahr\n- Ist Jahr bisher\n- Ist Ø Monat\n- Prognose Jahr\n- Abweichung\n- Ampel\n\nBitte ergänzen:\n\n- Row Click auf Kategorie\n- zeigt bestätigte Buchungen dieser Kategorie\n- Möglichkeit einzelne Buchung direkt umzukategorisieren\n- Budgetstatus aktualisiert sich danach\n\n## Arbeitspaket 8: Income / Einnahmen übersichtlicher machen\n\nBitte Income-Seite verbessern:\n\n- geplante Einnahmen\n- tatsächliche Einnahmen\n- Prognose\n- Abweichung\n- Status\n\nFalls noch keine Income-Kandidaten in Runtime vorhanden:\n- klar anzeigen: „Noch keine Einnahmen importiert/erfasst.“\n- keine Fehlermeldung\n- keine Fake-Daten\n\n## Arbeitspaket 9: Keine produktive Bulk-Buchung ohne Freigabe\n\nBitte ausdrücklich beachten:\n\n- Du darfst Test-/synthetische Kandidaten in Test-DBs bestätigen.\n- Du darfst in Runtime nur Einzel-UAT nutzen, falls nötig, und danach wieder bereinigen.\n- Du darfst keine echten 2026-Kandidaten massenhaft produktiv bestätigen.\n- Baue die Funktionalität, aber lasse die Entscheidung mir.\n\n## Arbeitspaket 10: Tests\n\nBitte Tests ergänzen für:\n\n### Review Dashboard\n- Summary Counts korrekt\n- Filter funktionieren\n- Sortierung funktioniert\n\n### Batch Review\n- Mehrfachauswahl\n- Kategorie für Auswahl setzen\n- Preview vor Confirm\n- Confirm erzeugt Transaktionen\n- keine Blindbuchung\n\n### Split\n- Split-Summe muss Hauptbetrag entsprechen\n- Split Confirm erzeugt auditierbare Buchung\n- falsche Summe blockiert\n\n### Kategorie aus Review\n- neue Kategorie im Review erstellen\n- Kandidat wird zugeordnet\n- Audit\n\n### Rule MVP\n- Regel erstellen\n- Regel deaktivieren\n- Regel testen\n- Regel auf Kandidaten als Vorschlag anwenden\n- keine produktive Buchung\n\n### Effektive Ausgaben\n- bestätigte Ausgaben erscheinen\n- Filter funktionieren\n- Row Click Detail\n- Kategorieänderung aktualisiert Budgetstatus\n\n### Budgetstatus\n- Row Click zeigt Buchungen\n- Umkategorisierung aktualisiert Ist-Werte\n\n### Safety\n- keine echten Dateien im Git\n- keine Runtime-DB im Git\n- Git-Safety sauber\n\n## Verifikation\n\nNach Umsetzung:\n\n- Python compile\n- pytest\n- npm test\n- npm run build\n- source secret scan\n- frontend build secret scan\n- Git-Safety\n- git diff --check\n- Commit\n- Push\n\n## Abschlussbericht\n\nBitte berichten:\n\n- Review Dashboard verbessert: ja/nein\n- Batch Review implementiert: ja/nein\n- Split-UI implementiert: ja/nein\n- Kategorie aus Review erstellbar: ja/nein\n- Rule Manager MVP-light implementiert: ja/nein\n- Effektive Ausgaben verbessert: ja/nein\n- Budgetstatus Row Click verbessert: ja/nein\n- Einkommen-Seite verbessert: ja/nein\n- Keine produktive Massenbestätigung: bestätigt\n- Tests\n- Git-Safety\n- Commit-Hash\n- Push-Status\n- offene Grenzen"
},
{
"role": "assistant",
"content": "",
"reasoning": "**Taking action on coding repo**\n\nI need to act because it looks like the coding repository is related to a finance system. I should load my skills about finance system architecture, perhaps by checking GitHub to see what’s available. Also, I need to check the current working directory, which is the hermes-agent. Inspecting that should help me understand how to use my skills effectively. Let's get started!",
"finish_reason": "tool_calls",
"reasoning_content": "**Taking action on coding repo**\n\nI need to act because it looks like the coding repository is related to a finance system. I should load my skills about finance system architecture, perhaps by checking GitHub to see what’s available. Also, I need to check the current working directory, which is the hermes-agent. Inspecting that should help me understand how to use my skills effectively. Let's get started!",
"codex_reasoning_items": [
{
"type": "reasoning",
"encrypted_content": "gAAAAABqCciDHMEDRkNfy549neJLo93FbBV3qD47w8NyAMEY9qF1q6G0gtsfGceKf4IgpLl4rVJFI4-EVCTnTK6EwQd-LjNSMlB9xHG64EZybUC-9xmctCOrT3HM_OsYEH4HgZG57ssFJaTko2cQdPEZ6C1wvSnrRqfnCTBGF4BzRaiQb-QvJLCh8MXcN4R4lkxm25R6y1x0OmVSDjoyPlYRVYdlL2aUSNbJgbXPqxZYFrf6zMh2z3bcpoDCe8b8PcnX_yI6rqtlNrSwQKnNXBrI1tR8EDO9qNEh3v6_QdQ7BBk5s3edAnrDMb3s1xiY2Tvo_IeBw6zQM_gClgYj_s1ygif79c5cn216hAYId5zdGPrPwM1dz5YKFP14zDQYtiP7fnPs7JLq5pNl04StoDqzsK7onrVdl1V795fYTjYmtlH6lORvyjHpHETSQmkEgfpE-mtH9jAhmk7xmUnuN-j3wUUr9q9jeYpicSgwtOAlGrs21nVburZnS44m9-LZYyk11UFZPDAeynQDaSWWVWIvO3UXXXb69JsSvBzNjHfIHL47l3OH-DSb4V43naI3OG2jmckIyvv4qlwZ4JN3m_mnNt7U4EomhMoSoUYIIkqSzcfLzAuiuUq80FccyxGTr86X_KAANaZrlMxr1gVNk_wQxsVP6mnHDd99p7hicZQr_nFdEjJJ_o3EAXQsEHqpHwWQRGvSDwT0uVqW7RE6egPAJce6oZNViHqMSoSt685nJsuYaQc4-YPVcRCnZBWJC8wyJoPkW9gCO8fI-Hf4eWf8DlyEnDOsv_YbU57iXBtWfqbddY8L8Tsz8QCXu1t_UvZulby7NdqgDzEzVXXQCaKUKaLgxcLJVQtEH0bNFzHF1psidD2OggECpMtpLV_csWrUWQSaX4SRYexN0mcWOt0n5p7o4lKyWuy-ajY1KIKEPnevEPgrADv0J62G-dP3SmpZ9W3CxXSpQSwZPRQlr6uDl7SBAvPcDPzDgTEncbelg5EJLtKB--GbM3TetQFXo74lmH99wYQvo_wjZoTcNFmlNk6622SccE5nOzr7qGFnHlk5NseuROYv7vuvNMBEmF4dHyKRQSmnbtFgn5bE36IMfOYtDfM7TtYdCvibG05yOO8MCfzUo06P_9hetjvCxaHuDv6E-S4m",
"id": "rs_09eaf407cd9fcc63016a09c880ac648191a7165b19b2ec1dec",
"summary": [
{
"type": "summary_text",
"text": "**Taking action on coding repo**\n\nI need to act because it looks like the coding repository is related to a finance system. I should load my skills about finance system architecture, perhaps by checking GitHub to see what’s available. Also, I need to check the current working directory, which is the hermes-agent. Inspecting that should help me understand how to use my skills effectively. Let's get started!"
}
]
}
],
"tool_calls": [
{
"id": "call_9y7KhTdFkxf5HfoGvIzkH05W",
"call_id": "call_9y7KhTdFkxf5HfoGvIzkH05W",
"response_item_id": "fc_09eaf407cd9fcc63016a09c88378348191a99c91ed758a5858",
"type": "function",
"function": {
"name": "skill_view",
"arguments": "{\"name\":\"finance-system-architecture\"}"
}
}
]
},
{
"role": "tool",
"name": "skill_view",
"content": "{\"success\": true, \"name\": \"finance-system-architecture\", \"description\": \"Design/implement the user's JARVIS Finance System: CHF ledger, portfolio accounting, dashboard, imports, reports, secrecy.\", \"tags\": [\"finance\", \"portfolio\", \"accounting\", \"dashboard\", \"csv\", \"fx\", \"tax\", \"privacy\"], \"related_skills\": [], \"content\": \"---\\nname: finance-system-architecture\\ndescription: \\\"Design/implement the user's JARVIS Finance System: CHF ledger, portfolio accounting, dashboard, imports, reports, secrecy.\\\"\\nversion: 1.0.0\\nauthor: Hermes Agent\\nlicense: MIT\\nmetadata:\\n hermes:\\n tags: [finance, portfolio, accounting, dashboard, csv, fx, tax, privacy]\\n---\\n\\n# Finance System Architecture\\n\\n## Important linked references\\n\\n- `references/vue-dashboard-tailscale-api-base.md` — FinanceManager Vue dashboard Tailscale restart/troubleshooting: set `VITE_API_BASE_URL` to the browser-visible Tailscale Vite origin, verify `/api/runtime/status` through Tailscale, and avoid remote-browser `127.0.0.1` confusion.\\n\\n## When to use\\n\\nUse this skill whenever the user asks about the **Finance System – Architektur & Scripts** topic or work involving the user's personal finance-management system: portfolio imports, transaction accounting, Bank/Broker/CSV processing, dashboards, reports, tax/income views, FX logic, market data jobs, alerts, or later implementation.\\n\\nDo **not** use it for pure investment-strategy discussion; those belong in the separate finance strategy topic unless the user explicitly asks for technical implementation.\\n\\n## Core principles\\n\\n1. **No automated trading.** The system may prepare orders and recommendations, but never executes orders or connects to broker trading APIs for execution.\\n2. **Accounting first, opinions later.** MVP must prioritize correct transaction ledger, cash ledger, FX handling, income/tax data, and dashboard before complex scores, news, analyst ratings, or LLM analysis.\\n3. **Rule-based first.** Calculations, scores, alerts, rebalancing and reports must be deterministic and auditable. LLMs may only summarize/explain text or assist extraction with human review.\\n4. **CHF as base currency.** Store every transaction in original currency and CHF using historical FX at transaction/payment date; current valuations use current FX.\\n5. **Financial data secrecy.** Never expose real financial values, portfolio files, raw exports, reports with real numbers, DBs, API keys, or credentials in chat, GitHub, logs, or shared docs unless the user explicitly approves scope.\\n6. **Data ownership.** All data must be exportable: CSV, JSON, Parquet, SQLite/Postgres dump, Markdown/HTML/PDF reports.\\n\\n## Current v0.3 decisions\\n\\nSee `references/spec-v0.2-decisions.md` for the baseline, `references/spec-v0.3-decisions.md` for the priority update, `references/spec-v0.4-implementation-blueprint.md` for the MVP 1 technical implementation blueprint, `references/spec-v0.4.1-corrections.md` for accepted corrections, `references/spec-v0.5-repo-blueprint.md` for the concrete repo/implementation plan, `references/mvp-safety-gate-implementation.md` for the implementation safety-gate sequence and verification commands, `references/mvp-import-pipeline-tasks-9-13.md` for the synthetic CSV import/FX/transaction pipeline pattern, `references/mvp-ledger-cash-wac-positions.md` for cash/WAC/position ledger calculation rules and tests, `references/mvp-crypto-wallet-ledger.md` for crypto wallet/assets/holdings/transfers/buy-sell/fee MVP rules, `references/crypto-xlsx-initial-snapshot-dry-run.md` for the controlled real-file XLSX crypto initial-snapshot dry-run pattern and mapping decisions, `references/crypto-xlsx-initial-snapshot-production-import.md` for productive safe-subset XLSX import operations, compatibility/partial-write recovery, and dashboard correction follow-up, `references/crypto-single-coin-runtime-test.md` for controlled one-coin productive runtime tests and the coin-vs-holding ambiguity guard, `references/multi-wallet-crypto-and-equity-manage-ui.md` for approved multi-wallet single-coin snapshots, SQLite Decimal/TEXT affinity pitfalls, and synthetic-first Equity/ETF Manage UI patterns, `references/crypto-price-refresh-and-idempotent-btc-runtime.md` for robust batch price refresh and re-run-safe BTC/single-coin runtime checks, `references/crypto-runtime-db-consistency-verification.md` for post-import runtime DB consistency checks, exact safe-set verification, dashboard no-live-API testing, and runtime-only report generation, `references/mvp-crypto-manual-management-ui.md` for audited dashboard add/correct/remove crypto holding workflows after an initial snapshot, `references/mvp-market-import-safety-hardening.md` for CoinGecko provider/backoff, CSV importer completion, alert lifecycle, Git-safety and crypto Decimal precision patterns, `references/git-safety-verification-coverage.md` for the verifier-warning workflow and minimum Git-safety test matrix, `references/mvp-streamlit-dashboard.md` for the safe local Streamlit dashboard shell/read-model/testing pattern, `references/mvp-crypto-report-export.md` for local-only crypto report context/export rules, `references/github-token-push-from-drive.md` for the safe Google-Drive-token push workflow when local FinanceManager commits are ahead of GitHub, `references/runtime-backup-and-broker-mapping.md` for runtime DB backup/restore hardening plus PostFinance/True Wealth/Raiffeisen structural mapping/dry-run rules, `references/import-wizard-mapping-layer.md` for the Phase B1 import wizard, instrument/account mapping, review queue, data-quality and read-only safety pattern, `references/broker-parser-dry-run-pipeline.md` for the Phase B2 synthetic broker parser, dry-run, runtime review-item and UI quality-flag pipeline, `references/controlled-real-broker-dry-run.md` for the Phase B3 controlled real-file dry-run pattern with backup, aggregate-only reporting, runtime-only review items, and no productive position writes, and `references/broker-review-queue-hardening.md` for Phase B4 manual review queue filters, auditable mapping actions, import-readiness gating, dry-run session management, and Data Quality Center hardening without productive portfolio writes, and `references/guided-manual-review-pilot.md` for Phase B5 small-scope runtime metadata pilots that move selected review items to `ready_for_import` without productive imports, `references/broker-import-execution-plan-mini-import.md` for Phase B6 final dry-run execution plans plus exactly-one productive broker mini-import, `references/broker-mini-import-regression-and-corrections.md` for post-mini-import regression, void/correction workflows, backup/restore checks, and one-at-a-time follow-up imports, and `references/broker-mini-import-completion-and-dq-consolidation.md` for completing the last ready True-Wealth mini-import plus active-vs-resolved Data Quality alert consolidation, `references/equity-fx-market-data-phase-c.md` for the local FX/market-data foundation after broker mini-imports and before broader Equity/ETF valuation/imports, and `references/true-wealth-valuation-pilot-c2.md` for the True-Wealth valuation pilot safeguards around hedge metadata, provider mappings, historical FX failures, timestamp alignment, corporate-action flags, and aggregate-only runtime reporting, `references/true-wealth-mapping-candidate-review-c3.md` for the controlled mapping-candidate review pattern before any productive Equity/ETF price or FX valuation, `references/instrument-catalog-manual-add-workflow.md` for the general Instrument Catalog + dashboard manual-add workflow that must underpin future Equity/ETF/Crypto adds, and `references/public-instrument-lookup-candidate-generator.md` for runtime-only public metadata lookups, candidate review packs, alert dedupe, and no-auto-selection rules, and `references/mapping-candidate-ranking-decision-pack.md` for ranking raw candidates into a manual-review decision aid, Data Quality alerts, confirmation workflow, and migration pitfalls, and `references/dashboard-user-mode-ux-and-runtime-review.md` for User/Admin dashboard navigation, crypto wallet drilldowns, audited Crypto Manage actions, and runtime-only mapping demotion patterns, and `references/dashboard-functional-ux-stabilization.md` for removing duplicate Streamlit navigation, hiding unfinished User Mode pages, enforcing Read-only/Edit Mode button behavior, and testing explicit-click price/report actions, and `references/dashboard-usability-simplification.md` for the stricter MVP User Mode reduction, calm Command Center, compact Crypto/Wallet pages, direct inline action forms, and disabled-not-fake button rule, and `references/crypto-adjusted-xlsx-and-equity-docx-dryrun.md` for adjusted real crypto XLSX completion using CoinGecko IDs plus aggregate-only True Wealth/PostFinance DOCX Equity/ETF dry-run preparation, and `references/equity-docx-instrument-resolver-review-workflow.md` for the controlled Equity/ETF DOCX staging, resolver, review queue, local-Qwen-as-parser-only, and confirmed-only initial-snapshot workflow, `references/manual-equity-etf-entry-wizard.md` for the preferred manual Equity/ETF search/add-position wizard when DOCX imports are too unsafe or not approved, and `references/manual-position-entry-provider-ux.md` for User Mode placement, explicit-click provider lookup, runtime-secret loading, caching, and final verification/push discipline, and `references/equity-etf-provider-integration-openfigi-fmp.md` for the concrete OpenFIGI-first/FMP-second integration pattern, provider merge rules, local-price preview semantics, and `update-equity-prices` workflow, `references/provider-integration-sprint-v2.md` for the multi-provider Free/Fallback pattern covering OpenFIGI/FMP/Finnhub/Twelve Data/Massive/EODHD, explicit-action dashboard semantics, sanitized status/error categories, and low-volume EODHD handling, and `references/provider-secret-auth-debugging.md` for targeted runtime-secret/provider-auth debugging when keys are present but OpenFIGI/FMP smoke tests fail, and `references/mvp-freeze-zwischenbericht-abnahme.md` for the MVP-freeze Zwischenbericht/Abnahmeplan pattern after an accepted implementation sprint, and `references/mvp-dashboard-workflow-regressions.md` for MVP dashboard regression coverage around cached FX, Cash wizard writes, explicit-click Equity/Crypto price refresh, Crypto cache-detail UX, runtime cleanup archives, and authenticated push verification, and `references/mvp-acceptance-dashboard-fx-valuation-cleanup.md` for acceptance-mode dashboard fixes covering hidden internal FX statuses, cache/provider/manual/missing FX flows, Equity/ETF valuation labels, runtime cleanup backup/audit, and final push verification, and `references/dashboard-user-mode-finalization-and-push.md` for integrated Portfolio User Mode finalization, hidden report paths, post-pytest Git-safety cleanup, Streamlit restart sanity, and authenticated push/hash verification, and `references/fastapi-vue-dashboard-transition.md` for the FastAPI read-only API + Vue User Mode transition pattern, DTO contracts, frontend Git-safety additions, and TestClient threading pitfall, `references/vue-ux-reference-sprint-pattern.md` for controlled UX-reference analysis/design-matrix/UI-system work on the read-only Vue vertical slice without copying external code/assets or expanding feature scope, and `references/vue-dashboard-autoload-runtime-operations.md` for auto-loading local FastAPI runtime DTOs, launch/browser sanity, SQLite `check_same_thread=False`, and final push verification, and `references/vue-cockpit-v11-preview-confirm-drawers.md` for the usable Vue cockpit pattern covering Preview/Confirm manual entry, detail drawers, roadmap links, stale-process browser sanity, frontend build location, changed-file secret scans, and authenticated remote-hash verification, `references/vue-cockpit-uat-preview-confirm-acceptance.md` for safe local Vue/FastAPI UAT of Cash/Equity Preview → Confirm → Audit flows, ledger-vs-compatibility writes, dry-run no-write verification, and the `JARVIS_FINANCE_DB_PATH` pytest contamination pitfall, and `references/crypto-vue-dashboard-readmodel-hardening.md` for Crypto Vue/FastAPI readmodel fixes around holdings-before-prices aggregation, compact crypto UX, sparkline empty states, wallet drilldowns, dashboard start/stop scripts, and full verification/push discipline, and `references/vue-manual-entry-v12-acceptance-completion.md` for completing Vue/FastAPI manual-entry acceptance flows including provider-search response DTOs, FX-aware previews, audited remove/account/wallet endpoints, review/test cleanup, cache/build cleanup before Git-safety, and authenticated remote-hash verification, and `references/vue-provider-preview-confirm-hardening.md` for provider-search/cross-listing dedupe, preview-as-read-only, confirm-as-write-boundary, cached CoinGecko URL sanitization, focused reviewer prompts, and post-test cleanup before Git-safety, and `references/vue-manual-position-v122-hardening.md` for v1.2.2-style manual-position provider warning normalization, candidate snapshot propagation, Equity status labels, compact Crypto layout, and remote-hash push verification. Key decisions:\\n\\n- Platform/account views: Raiffeisen, PostFinance, True Wealth, Crypto-Coins, plus Gesamtansicht.\\n- Existing Google Drive files under `Finanzen/01 Aktueller Stand Portfolios` are **initial-snapshot reality checks**, not full ledgers.\\n- MVP means the smallest useful first version: correct data foundation, ledger, CHF/FX logic, platform views, crypto inventory, watchlist, reports, workflows and audit log.\\n- Stock/ETF tax documents are downgraded: optional reference/control module and dividend/distribution transparency, not MVP 1 and not a complete tax dossier replacement.\\n- Crypto inventory is promoted to MVP 1: wallet/platform-level holdings, Crypto PDF, CoinGecko valuation, stale-price warnings and manual verification.\\n- Crypto snapshot values are stale; use quantities only and recalculate current values via CoinGecko later.\\n- MVP dashboard recommendation: Streamlit + SQLite + SQLAlchemy/Pandas/Plotly for the initial safe operations/admin shell; when moving toward a premium User Mode, use a thin FastAPI read-only API first and Vue 3 + TypeScript frontend later, following `references/fastapi-vue-dashboard-transition.md`.\\n- MVP inputs: manual dashboard entry, standard CSV import/export, later broker/bank export import. No direct broker integrations initially.\\n- Telegram crypto buys/sells/transfers should be parsed as pending transactions and only saved after explicit confirmation.\\n- Every portfolio or crypto change must create a History/Audit entry.\\n\\n## Budget-module PRD, source analysis and Phase-1 implementation\\n\\nWhen the user asks for a Budget/Cashflow module PRD based on Google Drive source material, follow `references/budget-module-prd-drive-analysis.md`: temp-only structural analysis of PRDs, XLS/XLSX budget workbooks and CSV samples; no real values in chat; no repo/raw-data commits; create both Markdown and Google Doc artifacts in Drive; verify IDs and cleanup temp files before reporting.\\n\\nWhen implementing the first Budget/Cashflow foundation in FinanceManager, follow `references/budget-phase1-manual-ledger-foundation.md`: build only the manual ledger foundation first (budget accounts/categories/tags/manual transactions/transfers/monthly overview), keep imports/rules/duplicates out of scope until separately authorized, use Preview → Confirm as the write boundary, store exact money/FX fields as Decimal text where precision matters, and run the backend/frontend/Git-safety cleanup gate before commit/push.\\n\\nWhen extending Budget/Cashflow into runtime category/tag administration, budget-plan seeds, or workbook-derived dry-run helpers, follow `references/budget-phase11-runtime-category-seed.md`: keep Preview → Confirm → Audit as the write boundary, archive/deactivate tags instead of hard-deleting, keep UI free of prominent technical IDs, handle real budget workbooks temp-only with aggregate reporting, support old `.xls` through optional `xlrd`, run frontend commands from `frontend/`, and verify with compile/full pytest/frontend tests/build/scoped Ruff/secret scan/remote-hash push.\\n\\nWhen moving workbook-derived budget seeds from dry-run into persistent review candidates and a Vue/FastAPI review workflow, follow `references/budget-phase12-excel-seed-review.md`: persist only `budget_seed_candidates` during seed, keep Confirm as the productive write boundary, expose unclean ranges for review, align persistent parser counts with the validated dry-run analyzer, verify `productive_budget_plan_mutation=false`, scan build artifacts before cleanup, then remove generated frontend/caches before Git-safety.\\n\\nFor Budget/Cashflow Phase 1.4 workviews, extend the same safety pattern: create a verified runtime DB backup before migration/seed, use schema migrations for review-only transaction candidates, never confirm CSV/Migros/VISA rows as productive transactions during seed, expose Import Review and Income Forecast as review/work views, keep row-click drawers backed by filtered API endpoints, allow category/transaction edits only through audited update endpoints, convert/reopen unclean candidates with audit, report aggregate counts only, run full backend pytest plus frontend vitest/build, remove `frontend/dist`, `.vite`, `node_modules`, `.pytest_cache`, and `__pycache__` before Git-safety, then push and verify remote hash.\\n\\nFor Budget/Cashflow source-aware import correction after Phase 1.4-style candidate floods, follow `references/budget-source-aware-import-review.md`: keep Migros CSV as the source of truth for Migros expenses, aggregate Migros article rows into receipt-level review candidates, mark overlapping credit-card Migros rows as `covered_by_migros`, keep Galaxus review-required/splittable, classify AKB/Raiffeisen income and own-account transfers without booking internal transfers as income/expense, preserve Dry run → Review → Confirm, and report only aggregate counts/statuses.\\n\\nFor Budget/Cashflow Phase 1.6-style user category rules over existing import-review candidates, follow `references/budget-phase16-user-category-rules.md`: create/extend explainable candidate metadata such as `rule_id`, `rule_name`, `classification`, `review_reason`, confidence and `covered_by_source`; re-evaluate only open scoped candidates after a verified runtime backup; keep Confirm as the only productive write boundary; expose review tabs for subscriptions, Galaxus/Digitec, Migros threshold cases, health, transport, shopping, pets, admin/unclear and covered-by-Migros; and verify aggregate-only that productive budget transaction count did not change.\\n\\nFor Budget/Cashflow Phase 1.7-style UX/workflow refactors, follow `references/budget-phase17-ux-workflow-refactor.md`: separate Import Review, Category Assignment, Effective Expenses, Category Tree, Budget Planning and Budget-vs-Actual into distinct work surfaces; replace generic read-only messaging on work pages with Review-/Arbeitsmodus; make every visible action functional or explicitly disabled; show local amounts in the UI but never in chat reports; keep candidates as candidates until Confirm; ensure category/transaction/candidate changes are audited; build category merge as Preview → Confirm; run full backend/frontend/Git-safety verification before commit/push.\\n\\nFor Budget/Cashflow Phase 1.8-style core-flow corrections, follow `references/budget-phase18-core-confirm-flow.md`: remove duplicated candidate work pages, make **Buchungen prüfen** the single central review page, require Confirm to create a real `budget_transactions` row with `source_type='import_candidate'`/`source_candidate_id`, persist `confirmed_transaction_id` on the candidate, show Effective Expenses only from confirmed expense transactions, include actual-only categories in Budgetstatus even without plan rows, archive UAT/test runtime artefacts after a backup and audit event, and check SQLite migrations for lingering `__phase*` FK references after table rebuilds.\\n\\n## Required data model areas\\n\\nAt minimum design for:\\n\\n- `platforms`, `accounts`, `instruments`\\n- `transactions` with original currency and CHF fields\\n- `cash_ledger`\\n- derived `positions_current`\\n- `dividends_distributions`\\n- `market_prices`, `fx_rates`\\n- `watchlist_items`\\n- `alerts`\\n- `decision_journal`\\n- `reports`\\n- `data_quality_issues`\\n\\n## CSV/import rules\\n\\n- For broker/bank imports (PostFinance, True Wealth, Raiffeisen), build the safe review layer before productive imports: Import Wizard, instrument mappings, platform/account mappings, dry-run summaries, Manual Review Queue, Data Quality Center and visible read-only/write-enabled state. Follow `references/import-wizard-mapping-layer.md`. When wiring parser output into the wizard, use the synthetic-first parser/dry-run/review-item pipeline in `references/broker-parser-dry-run-pipeline.md`. After explicit approval for real-file dry-runs, follow `references/controlled-real-broker-dry-run.md`: preflight, verified runtime backup, temp-only file handling, aggregate-only reporting, runtime-only review items, and verification that productive positions/ledger rows stayed unchanged. For DOCX portfolio extracts from True Wealth/PostFinance, follow `references/crypto-adjusted-xlsx-and-equity-docx-dryrun.md`: use structural/masked parsing only, count field coverage and mapping status, treat ticker-without-exchange as review-needed, and stop before productive import. If the DOCX dry run shows low ISIN coverage/no cost basis/no transaction history or the user withholds final import approval, switch to `references/equity-docx-instrument-resolver-review-workflow.md`: create `instrument_import_candidates`-style staging, use local Qwen only as an assistive parser if available, classify ticker-only rows as `needs_manual_review`, expose a Mapping Review Queue, and import only later from `confirmed` candidates after explicit approval. If the user rejects productive DOCX import or wants direct manual control, follow `references/manual-equity-etf-entry-wizard.md`: make the main path an audited portfolio-app-style search/add-position wizard with explicit candidate selection, no live API on render, no ticker-only auto-selection, hidden technical IDs in User Mode, cost-basis/FX quality alerts, and audit-log writes. When the user asks to make real dry-run results manually editable, follow `references/broker-review-queue-hardening.md`: add filters/detail views, auditable review-metadata actions, per-item/session import-readiness status, dry-run session management, and Data Quality Center counts while keeping productive portfolio writes disabled. For a guided manual-review pilot on real runtime items, follow `references/guided-manual-review-pilot.md`: select 1–3 clear ETF/equity items (prefer True Wealth), write only audited review metadata, correct only presence booleans if needed, derive `ready_for_import` via readiness logic, and verify productive tables remain unchanged. Before the first productive broker position import, follow `references/broker-import-execution-plan-mini-import.md`: create runtime-only execution plans from the ready review items, validate exact Decimal-text payloads from the runtime source, preserve missing FX/cost-basis quality flags, then import exactly one ledger `initial_position_snapshot` after a verified backup; do not write `positions_snapshot` as the source of truth. After the first mini-import and before broader imports, follow `references/broker-mini-import-regression-and-corrections.md`: verify import links/audit/Decimal handling/re-import blocking, add or test safe void/correction flows, run backup/restore-to-test-DB regression, and only then optionally import one more ready plan. When the user approves finishing a remaining small set, follow `references/broker-mini-import-completion-and-dq-consolidation.md`: first disambiguate active vs resolved stale crypto alerts with `check-data-quality --scope crypto --resolve-fixed`, then import exactly one remaining ready execution plan after backup/verify, and consolidate Review Queue/Data Quality counts without exposing values.\\n- Do not pivot into later analytics modules while the user has asked for import/mapping hardening. Monte Carlo, backtesting, rebalancing, scores, heatmaps and projections are later phases unless explicitly re-authorized.\\n- Use a canonical CSV format before broker-specific importers.\\n- Imports must be idempotent via external transaction ID or row hash.\\n- Importers must separate dry-run and commit modes, collect validation errors explicitly, and always create an `import_sessions` record even for dry runs or failed imports.\\n- Import results should classify rows as new, existing, duplicate, or failed; do not rely on `INSERT OR IGNORE` as the only duplicate-detection mechanism.\\n- Row hashes must be stable and deterministic: normalize key order, whitespace, and `None` versus empty strings while leaving domain-specific canonicalization to the importer.\\n- If a source file has only current holdings and no full history, import it as `initial_position_snapshot` / `initial_cash_snapshot`, not fake historical buys.\\n- Missing historic FX, fees, taxes, dates, investment case, target weights, or dividend details must be flagged as data-quality gaps.\\n\\n## FX and return rules\\n\\nFor acceptance-mode dashboard fixes around manual position entry and valuation UX, follow `references/mvp-acceptance-dashboard-fx-valuation-cleanup.md`: hide internal FX status codes from User Mode, treat CHF as `not_needed`, resolve USD/EUR via cache/provider on explicit save, require a conscious manual-or-missing choice when FX cannot be resolved, and never store `ok` without a rate. For real Runtime-DB Streamlit regression sprints spanning Alerts, navigation, FX/Cash, Equity/ETF interaction, cleanup, no-live-render, browser sanity, and final push verification, also follow `references/dashboard-real-ui-acceptance-regression.md`.\\n\\nFor foreign currency assets, always distinguish:\\n\\n- price/security performance in original currency\\n- FX effect to CHF\\n- realized P&L\\n- unrealized P&L\\n- dividends/distributions/income\\n- fees and taxes\\n- Total Return in CHF\\n\\nEvery transaction/income event needs:\\n\\n- original amount and currency\\n- historical FX rate to CHF, or explicit `fx_status=missing` with a critical data-quality alert\\n- FX source and timestamp\\n- CHF equivalent at event date when FX is known\\n\\nManual FX overrides require a non-empty note and an audit-log entry.\\n\\n## Ledger calculation rules\\n\\n- Cash after system start must be calculated from `transactions`; `cash_balances` is only a snapshot/cache/control view.\\n- `initial_cash_snapshot` and `initial_position_snapshot` are explicit starting truths, not fabricated historical trades.\\n- Implement Weighted Average Cost as the MVP internal performance method: buy fees increase cost basis, partial sells remove proportional cost basis, realized P&L is calculated against average cost, full sells zero remaining quantity/cost basis, and dividends do not alter quantity or cost basis.\\n- WAC is not tax-complete; do not add FIFO/LIFO/tax-lot logic in MVP 1, but keep transaction granularity so it can be added later.\\n- Position calculations must keep zero-quantity historical positions visible when sale history exists.\\n- Missing market price or FX must degrade data quality and suppress precise Total Return CHF rather than inventing a value.\\n- Before productive dashboard use, add alert deduplication per rule/entity/date.\\n- Market-FX for non-CHF market prices belongs to the later Market/FX provider block.\\n\\n## Crypto MVP rules\\n\\n- Crypto inventory is MVP-critical but must remain dummy-first until mapper/dry-run layers are ready; do not touch real coin/wallet files during core implementation. After explicit user approval, real files may be read only from temp/runtime for controlled dry runs using `references/crypto-xlsx-initial-snapshot-dry-run.md`; report only aggregate counts/labels and never quantities or values. Productive safe-subset imports require a second explicit approval and must follow `references/crypto-xlsx-initial-snapshot-production-import.md`: external runtime DB only, no real CSV/XLSX/reports/DBs in Git, Git-safety before/after, import-session + audit-log creation, and aggregate-only reporting. When the user supplies an adjusted real crypto workbook with CoinGecko IDs for missing/previously skipped assets, follow `references/crypto-adjusted-xlsx-and-equity-docx-dryrun.md`: detect the actual structure first, validate IDs against CoinGecko metadata, import only clear missing asset/holding subsets after verified backup, preserve Decimal/Text quantities, ignore legacy CHF values, and delete temporary source files.\\n- Wallet names must be unique; wallet addresses are optional and must not appear in tests/fixtures.\\n- Validate wallet type against: Hardware Wallet, Software Wallet, Exchange, Bank/Broker, DeFi, Sonstiges.\\n- CoinGecko ID is strongly recommended but not hard-required; missing ID creates a data-quality warning and blocks reliable valuation.\\n- Crypto symbols are not globally unique; symbol conflicts require warning/manual selection, never silent merging.\\n- Initial crypto holdings are confirmed snapshots by wallet+asset. Store legacy values for provenance/control only, never current valuation.\\n- Crypto holdings should be calculated/controlled from initial snapshots plus crypto transactions, not silently overwritten.\\n- Crypto transfers require different source/target wallets, quantity > 0, and same asset; coin fees reduce source-wallet balance.\\n- Negative wallet balances should be blocked in MVP and produce a critical alert unless a future explicit correction workflow is implemented.\\n- Crypto buys/sells with fiat impact must create/reference general ledger transactions via `source_type='crypto'`/`source_id` so cash remains ledger-consistent.\\n- Fiat fees should be ledger fees when unambiguous; avoid double counting when a fee is already embedded in sell net proceeds.\\n- Unknown fee currency creates a data-quality warning instead of invented accounting.\\n- Crypto Cost Basis/WAC is not required for MVP 1; MVP 1 prioritizes crypto holdings, wallet structure, current valuation, reports and audit.\\n- Crypto report context/export must be renderer-neutral and local-only: build from SQLite, aggregate by coin and wallet, use local cached prices only, preserve Decimal/Text values, ignore legacy snapshot values for current valuation, include data-quality warnings, and write generated files only to the external runtime reports directory.\\n- Crypto PDF export should be optional-dependency tolerant: prefer PDF when a renderer is available, otherwise fall back to HTML/Markdown with a clear warning while still recording report metadata and audit.\\n- After productive real-data crypto imports, run the post-import runtime verification in `references/crypto-runtime-db-consistency-verification.md`: exact safe-set membership checks, aggregate-only counts, audit/import-session traceability, dashboard render with no live API, runtime-only report generation, and Git-safety before/after.\\n- Manual dashboard additions/corrections/removals after an initial snapshot must follow `references/mvp-crypto-manual-management-ui.md`: no silent `crypto_holdings` mutation, explicit confirmation, mandatory notes for adjustments/removals, audit-log entries with old/new values, Decimal/TEXT quantity handling, no negative holdings, and no live API lookup during render. For dashboard UX/read-model work, also follow `references/dashboard-user-mode-ux-and-runtime-review.md`: split User Mode from Admin/Review/Debug, aggregate coins across wallets, hide technical IDs in normal views, keep cached-price semantics explicit, and make Crypto Manage actions audited workflows rather than raw table edits. For Vue/FastAPI Crypto page hardening, follow `references/crypto-vue-dashboard-readmodel-hardening.md`: aggregate holdings by asset/wallet before applying the latest price, test duplicate-price-row regressions, keep crypto UI compact, hide weak sparklines behind an empty state, and make wallet rows drill into wallet detail.\\n- Crypto tax logic, tax lots, FIFO/LIFO and detailed cost basis are later modules.\\n- Keep crypto assets separate from `instruments` for MVP 1, but later evaluate a shared Asset Master if it improves reporting, performance views or watchlists.\\n\\n## Income/tax rules\\n\\nFor dividends and ETF distributions, separately store:\\n\\n- gross dividend/distribution\\n- Swiss withholding tax 35%\\n- foreign withholding tax\\n- other tax/withholding if present\\n- net amount\\n- original currency\\n- historic FX to CHF\\n- gross/net CHF equivalents\\n- platform/account\\n- ISIN/ticker\\n- payment date\\n- ex-dividend date if available\\n\\nPlan later ESTV/ICTax support for Swiss year-end values and tax reports, but label outputs as technical reports, not tax advice.\\n\\n## Market-data rules\\n\\n- Dashboard reads primarily from local DB, not directly from APIs on page load.\\n- For Equity/ETF FX and market-data hardening after broker mini-imports, follow `references/equity-fx-market-data-phase-c.md`: validate active `missing_fx` alerts by original transaction currency first, correct only CHF false positives with audit, keep foreign-currency missing-FX alerts active until local historical rates exist, use ISIN as instrument identity and provider symbols only as market-data mappings, and run market-price providers only from explicit CLI/scheduled jobs.\\n- For the True-Wealth valuation pilot, follow `references/true-wealth-valuation-pilot-c2.md`: do not import more positions, do not guess provider symbols or hedge status, extend instrument metadata for currency hedging/status/valuation policy/corporate-action state, handle historical-FX paywalls as `manual_override_required` without crashing, align price/FX timestamps before claiming valuation precision, and report only aggregate runtime counts.\\n- For True-Wealth mapping-candidate review after the valuation pilot, follow `references/true-wealth-mapping-candidate-review-c3.md`: create runtime-only provider-symbol candidates keyed by ISIN/instrument, keep name-only or ticker-without-exchange candidates review-required, mark multiple candidates `needs_manual_review`, do not auto-select final mappings on ambiguity, keep hedge/instrument status unknown until confirmed, and ensure market-price dry runs call no provider when there are zero confirmed mappings. For public metadata lookup implementation details, provider result caps, OpenFIGI/Yahoo evidence handling, alert dedupe, runtime-only review templates, and UI/test requirements, use `references/public-instrument-lookup-candidate-generator.md`. For ranking/scoring candidates, generating a runtime-only user decision pack, surfacing `candidate_review_required` and same-ISIN multiple-listing risks, and confirming/rejecting candidates with audit logs, use `references/mapping-candidate-ranking-decision-pack.md`.\\n- For the broader Instrument Catalog + manual-add workflow, follow `references/instrument-catalog-manual-add-workflow.md`: documents/import files are initial-population aids, but the dashboard must allow audited manual add of equities/ETFs and later coins via search, candidate selection, account/wallet selection, Decimal/Text quantity, date, cost-basis/snapshot info and notes. ISIN is primary for equities/ETFs but can have multiple listings; ticker/symbol alone is never unique; provider symbols are market-data mappings only; CoinGecko ID is primary for crypto provider identity. External lookups may use only non-confidential instrument metadata and must never send quantities, balances, account/wallet IDs, cost basis, private notes or source documents. If DOCX/Broker import is not approved or structurally weak, use `references/manual-equity-etf-entry-wizard.md` as the preferred user path: search on explicit click, show candidates, require conscious selection, allow warned manual creation, then review/confirm an audited initial snapshot or transaction. For the concrete Dashboard/User-Mode implementation pattern, provider order (local/OpenFIGI/FMP/Finnhub/manual), runtime `.env` secret lookup, provider-result caching, and final compile/test/Git-safety/commit/push discipline, also follow `references/manual-position-entry-provider-ux.md` and `references/equity-etf-provider-integration-openfigi-fmp.md`. For multi-provider hardening across OpenFIGI/FMP/Finnhub/Twelve Data/Massive/EODHD, follow `references/provider-integration-sprint-v2.md`: support all documented secret aliases including `TWELVEDATA_API_KEY`, keep FMP primary, use Twelve/Finnhub/Massive as fallback, keep EODHD explicit low-volume only, sanitize status/error output, and ensure dashboard status uses non-probing render mode. If provider smoke tests report missing candidates or unreachable providers despite keys being stored, first follow `references/provider-secret-auth-debugging.md`: verify runtime secret loading without printing values, classify 401/403 as `auth_failed` rather than generic reachability failure, and improve status diagnostics before adding features.\\n- First Streamlit MVP should be a safe, local, read-first shell over SQLite with explicit synthetic demo DB creation only; page render functions must not call CoinGecko/market APIs, and tests should prove that by monkeypatching provider methods to fail. When the user asks to move beyond Streamlit, do not mutate the working Streamlit UI first; add a small FastAPI read-only boundary with DTOs and tests, then build a Vue vertical slice. Follow `references/fastapi-vue-dashboard-transition.md`.\\n- Functional UX stabilization work must prioritize making the existing dashboard genuinely usable before adding more pages. Follow `references/dashboard-functional-ux-stabilization.md`: use one navigation system, move executable page modules out of Streamlit `pages/` when using a custom router, hide unfinished pages from User Mode, replace text-link/fake actions with real buttons or disabled controls, enforce session-based Edit Mode for writes, keep IDs/debug notes out of User Mode, and test price/report provider calls happen only on explicit clicks. If the user says the dashboard is still too busy or prototype-like, follow `references/dashboard-usability-simplification.md`: reduce User Mode to only currently useful pages, keep the Command Center calm, avoid misleading `CHF 0.00` Equity/ETF displays when valuation is incomplete, make Crypto the primary MVP surface, and ensure every visible button either works immediately, opens an inline form, or is disabled with a reason. For regression hardening after manual-position/provider work, also follow `references/mvp-dashboard-workflow-regressions.md`: cache-first historical FX, audited Cash wizard writes, explicit-click Equity/ETF and Crypto refreshes, no live API on render, User Mode cache-source labels without raw IDs, and manifest-based runtime cleanup.\\n- Use scheduled jobs to update prices/FX/news.\\n- Store provider, timestamp, quality status and freshness for every price/FX point.\\n- Implement caching, local historical storage, batch requests where possible, provider rate-limit queues, and backoff on HTTP 429.\\n- CoinGecko integrations must be mockable and unit-tested without real API calls; treat HTTP 429 as non-fatal and return a non-fresh data-quality status plus alert after configured retries.\\n- Crypto price refresh belongs in CLI/scheduled-job commands, not dashboard/report rendering. Prefer CoinGecko batch requests over per-coin requests; expose conservative controls (`--sleep-seconds`, `--max-retries`, `--initial-backoff`, `--max-backoff`, `--only-missing`, `--only-stale`, `--only-symbol`, `--limit`, `--dry-run`, `--currency`). Commands should use aggregate-only output (`total_assets/updated/cached/skipped/stale/warnings/errors/assets_still_missing_local_price`), cache by configurable `max_age_seconds`, store exact Decimal prices as TEXT in `crypto_prices`, and verify dashboard/report contexts read local prices with `live_api_calls=False`. For details and tests see `references/crypto-price-refresh-and-idempotent-btc-runtime.md`.\\n- After an accepted sprint, respect an explicit MVP freeze: do not add providers or large features; write/update the sanitized repo report and acceptance checklist, verify with compile/tests/Git-safety, commit/push, and prepare the next real end-to-end runtime test. See `references/mvp-freeze-zwischenbericht-abnahme.md`.\\n- Candidate providers: Finnhub, Financial Modeling Prep, Twelve Data, Stooq, Alpha Vantage selectively, Yahoo Finance fallback only; CoinGecko for crypto; later DefiLlama; ESTV/ICTax for Swiss tax values.\\n\\n## Crypto Decimal precision rules\\n\\nSQLite has no exact native DECIMAL. In the crypto context, store quantities, prices, crypto fees and valuation-relevant amounts as exact Decimal strings (`TEXT`) and compute with Python `Decimal`. Never convert crypto quantity/price/fee/valuation through float/REAL. Format Decimal writes with `format(value, \\\"f\\\")` so small quantities such as `0.00000001` do not become scientific notation. CSV importers must validate Decimal strings, and tests should assert high-precision preservation plus SQLite text storage. For multi-wallet real-data snapshots and Equity/ETF ledger fields, also see `references/multi-wallet-crypto-and-equity-manage-ui.md`; SQLite `NUMERIC` affinity can coerce Decimal-looking text to `real`, so schema/migrations should use `TEXT` for exact quantities, FX rates, original amounts and CHF equivalents where precision matters.\\n\\n## Alert lifecycle rules\\n\\nAlerts must be deduplicated and lifecycle-aware before dashboards run repeated calculations. Use a dedup key from `rule_id`, `entity_type`, `entity_id`, `priority` and an error fingerprint. Alerts should carry `status` (`active`, `resolved`, `muted`), `resolved_at`, optional `muted_until`, `last_seen_at`, `occurrence_count`, `fingerprint` and `dedup_key`. Re-seeing an identical active/resolved/muted alert should update `last_seen_at`/`occurrence_count` rather than create a new active row; changed fingerprints may create a new alert.\\n\\n## Git/GitHub safety\\n\\nCode/scripts/docs may be stored in a private Git repo, but never commit financial data. Use the `github-repo-management` skill and its `references/sensitive-code-backup.md` guidance.\\n\\nMinimum `.gitignore` for finance repos must block `data/`, `reports/`, `exports/`, `backups/`, DB files, spreadsheets, PDFs, DOCX, raw CSV/JSON, `.env`, tokens, credentials and client secrets. Only synthetic examples under `examples/` are acceptable.\\n\\n## MVP implementation safety gate\\n\\nBefore any real broker/bank/portfolio/crypto data is imported, build and verify the safety foundation described in `references/mvp-safety-gate-implementation.md`: repo skeleton, `.gitignore`, `.env.example`, runtime outside repo, settings validation, SQLite schema, synthetic fixtures, audit log, CSV dry-run, duplicate protection, and Git-safety scan. Run the safety scan before staging, committing, or pushing. If GitHub credentials are needed from Drive, use them only in a temporary location and never echo token values.\\n\\n- Before broader productive imports, implement or verify the runtime backup loop in `references/runtime-backup-and-broker-mapping.md`: backup outside Git, SHA256 verification, restore requiring explicit confirmation, tests for backup/restore/checksum, and Git-safety coverage that blocks backup artifacts in the repo. For PostFinance/True Wealth/Raiffeisen, perform structural mapping and dry-runs only until the user approves import; report aggregate counts/field coverage/quality flags only, never real values or raw rows. If backup CLI output is too terse to evidence the backup path/checksum, call `jarvis_finance.runtime.backup.backup_runtime_db(...)` and `verify_backup(...)` directly and report only path-existence/outside-repo/checksum status, not DB contents.\\n\\n## Common pitfalls\\n\\n- Starting later analytics (Monte Carlo, backtesting, rebalancing, scoring, heatmaps, projections) while import/mapping/data-quality phases are still being hardened or while the user explicitly asked to postpone them.\\n- Letting the Budget/Cashflow foundation become a hidden import/rule-engine sprint. Phase 1 should be manual accounts/categories/tags/transactions/transfers/overview only, with Preview → Confirm writes and no VISA/Migros/Cumulus import, duplicate engine, or rule engine unless the user explicitly opens that next phase. See `references/budget-phase1-manual-ledger-foundation.md`.\\n- Treating workbook-derived Budget/Cashflow seeds as safe productive imports. Excel seed parsing may persist review candidates, but must not mutate productive budget plans until explicit Confirm. Always compare candidate counts with the validated dry-run analyzer, preserve ambiguous rows as unclean ranges, and verify aggregate-only that productive budget table counts did not change during seed. See `references/budget-phase12-excel-seed-review.md`.\\n- Letting Phase 1.5-style source-aware Budget/Cashflow import corrections remain row-granular. Migros/Cumulus article rows are detail data, not normal review candidates; aggregate them to receipts and mark previous article candidates `superseded`/archived with audit after a verified runtime backup. Credit-card Migros rows must be `covered_by_migros`, Galaxus must stay review-required/splittable, and AKB/Raiffeisen own-account transfers must be transfer candidates rather than income/expense. See `references/budget-source-aware-import-review.md`.\\n- Over-broad merchant matching in Budget user-category rules. A plain pattern such as `M ` can falsely classify unrelated merchants as Migros; use normalized explicit merchant tokens, guard special abbreviations, and exclude `Migrol` from grocery/Migros coverage when it should classify as fuel/transport. User-rule passes must update candidate metadata only, keep Galaxus/Digitec review-required/splittable, and prove no productive transactions were created. See `references/budget-phase16-user-category-rules.md`.\\n- Treating a holdings snapshot as a real transaction history.\\n- Mixing price P&L and FX P&L.\\n- Letting dashboard refreshes hammer APIs.\\n- Committing realistic sample data to Git because the repo is private. Private GitHub is not a financial data vault.\\n- Showing real portfolio values back to the user unnecessarily in summaries.\\n- Designing Git-safety scans that block source package names like `src/.../imports` or `src/.../reports`; block top-level runtime/data directories instead.\\n- In broker/bank real-file dry-runs, aggregating Data Quality flags only for `review_status='open'`. Blocked review items are still actionable data-quality issues; include `open` and `blocked` so defensive Raiffeisen aggregate-only rows remain visible.\\n- Letting manual review actions blur into productive import actions. Review-metadata writes (ISIN/ticker/exchange/account/instrument confirmations, ignore/resolve, dry-run archive/current status) may be allowed with audit logs, but they must not write `positions_snapshot`, `transactions`, `cash_balances`, ledger rows, or cash snapshots. Name-only mapping must never become `mapped`/`ready_for_import`; ticker-only without exchange remains review-needed; `aggregate_only` remains blocked for instrument import. In guided pilots, do not force `ready_for_import` by direct status edits; let readiness logic derive it after audited metadata actions, and if parser presence flags were missed, correct only boolean presence metadata with an audit event, never actual quantities/values.\\n- Skipping the execution-plan layer between ready review items and productive broker imports. A final dry-run should materialize runtime-only execution plans with source row hash, target IDs, exact Decimal-text payload summary, quality flags and planned write status. Productive mini-imports should consume exactly one planned ready payload, write exactly one ledger `initial_position_snapshot`, mark that plan/review item imported, leave the other ready plans untouched, and verify zero direct `positions_snapshot`/cash writes. After the first productive mini-import, do not proceed straight to mass import: run regression checks, implement/test void and correction audit paths, verify backup/restore to a separate test DB, and import at most one additional ready plan. When later completing the final ready plan, first resolve Data Quality alert ambiguity by counting only active alerts (`status='active'`) and treating resolved rows as history; then import exactly one plan and consolidate counts. See `references/broker-import-execution-plan-mini-import.md`, `references/broker-mini-import-regression-and-corrections.md`, and `references/broker-mini-import-completion-and-dq-consolidation.md`.\\n- Running compile/tests before Git-safety and forgetting generated `__pycache__`; remove or ignore generated caches before safety scans. `pytest` may recreate `.pytest_cache`, so remove `__pycache__`/`.pytest_cache` again after test execution and before `git-safety-scan`.\\n- Treating every active `missing_fx` alert after a broker mini-import as a bug. First aggregate by original currency: CHF rows may be false positives (`fx=1`, `not_needed`, resolve with audit), but foreign-currency rows without local historical FX should remain active and block precise CHF valuation until rates are populated.\\n- Auto-mapping Equity/ETF market prices by instrument name or ticker alone. ISIN is the instrument identity, but one ISIN can have multiple listings across exchange/currency, so even ISIN matches may need candidate selection. Provider symbols are separate market-data mappings, ticker without exchange is review-needed, and missing provider symbols should create deduplicated Data Quality alerts rather than triggering provider guesses. For mapping-candidate phases, candidate creation is allowed in the runtime DB, but final selection must remain manual/audited unless the user explicitly approves and the match is unambiguous. Public lookup endpoints can return dozens of listings; cap result sets, dedupe active mapping alerts by rule/entity, and do not treat search-fallback results as ISIN-confirmed unless the provider response supplies ISIN evidence.\\n- Designing future portfolio entry as import-file-only. Imports/documents are for initial population and existing holdings; the system must also support dashboard manual add/search workflows with audited ledger writes, explicit candidate selection, Decimal/Text quantities, required notes where history is incomplete, and no silent overwrites. If the user says DOCX/Broker files are too unsafe or not approved, stop the productive import path and pivot to the manual Equity/ETF wizard in `references/manual-equity-etf-entry-wizard.md`; keep the review queue optional/admin-only, ensure manual-entry broker accounts exist, and do not expose technical IDs in User Mode. The manual entry path must be visible in normal User Mode (e.g. `Position hinzufügen`) with Command Center routing; provider lookup must be explicit-click only, cache results locally, load provider keys from runtime secrets, and degrade missing keys to warnings. See `references/instrument-catalog-manual-add-workflow.md` and `references/manual-position-entry-provider-ux.md`.\\n- Letting an Equity/ETF market-price dry run call providers when no confirmed mappings exist. The safe outcome is `total_mappings=0`, no provider calls, and aggregate reporting only.\\n- For valuation pilots on historical broker snapshots, treating a computable current market value as a clean performance result. `snapshot_only`, `cost_basis_uncertain`, missing hedge confirmation, missing historical FX, stale price/FX timestamp alignment, or suspected corporate actions must keep valuation `partially_valuable` or suppress precise Total Return until reviewed.\\n- Inferring CHF-hedged ETF behavior from the name alone. Hedge status is methodology-critical: unknown status should create a warning/review item; confirmed CHF hedging must prevent naive free FX-PnL attribution.\\n- Crashing or blocking on historical-FX provider paywalls. Persist `manual_override_required`, create deduplicated alerts, keep the flow moving, and require a noted audited manual override instead of guessing latest FX.\\n- Productive real-data imports can partially write before surfacing schema/importer compatibility issues. Do not rerun blindly or dump row-level data; inspect aggregate counts and unique identifiers needed for idempotency, patch/complete via missing-only inserts, then verify committed import session, audit log, Git clean and Git-safety OK.\\n- In controlled one-coin runtime tests from a real XLSX, distinguish **one coin/asset** from **one holding/wallet row**. A coin may be split across multiple wallet columns; if the user required an unambiguous target wallet or one holding, stop and ask rather than choosing a wallet silently. If the user explicitly approves multi-wallet import for that coin, import exactly one asset plus one audited holding per non-empty approved wallet cell; verify aggregate counts only. If re-running a BTC/single-coin test finds the exact approved asset and holdings already present, do not import duplicates; perform the aggregate-only idempotency checks and report that asset/holding counts did not increase because duplicate prevention succeeded. See `references/crypto-single-coin-runtime-test.md`, `references/multi-wallet-crypto-and-equity-manage-ui.md`, and `references/crypto-price-refresh-and-idempotent-btc-runtime.md`.\\n- After productive crypto imports, do a separate aggregate-only consistency pass before building new features: DB/runtime outside repo, exact safe-subset membership, no duplicate/negative/orphan holdings, correct snapshot date, null legacy-value fields when ignored, import sessions committed, holding-level audit traces present, dashboard local-read/no-live-API verified, and report generated only under runtime reports.\\n- Letting a dashboard correction UI mutate crypto holdings directly. Add/remove/edit operations must become audited transactions or `manual_adjustment` events with mandatory notes where appropriate; preserve history instead of silently overwriting balances.\\n- Letting normal dashboard pages become technical raw-table dumps. For this user's Finance dashboard, implement an Alltag/User Mode that hides internal IDs/provider internals and presents aggregate, actionable read models; keep raw identifiers, candidate details and audit/debug fields in Admin/Review/Data Quality views. See `references/dashboard-user-mode-ux-and-runtime-review.md`.\\n- Treating Streamlit/dashboard UX stabilization as a chance to build more modules. When the user complains about prototype feel or duplicated workflows, first eliminate duplicate navigation, empty User Mode pages, fake/text buttons, unclear Read-only/Edit state, and raw technical IDs. Keep unfinished pages in Admin/Debug or hide them; every visible button must work, prefill a real workflow, or be disabled with a reason. For Budget candidate workflows specifically, do not keep Import Review and Category Assignment as two pages over the same rows; centralize them as **Buchungen prüfen** and ensure Confirm really books a productive transaction before claiming acceptance. See `references/dashboard-functional-ux-stabilization.md`, `references/dashboard-usability-simplification.md`, and `references/budget-phase18-core-confirm-flow.md`.\\n- Letting an accepted sprint half-finished because the next feature is tempting. If cleanup, final tests, Git-safety, commit, and push are still missing, finish those in order before starting new work. Remove `.pytest_cache`/`__pycache__` before and after tests, then run Git-safety, `git diff --check`, verify clean status, commit, push, and fetch/update `origin/main` to prove sync. If normal `origin` verification is unauthenticated but a FinanceManager runtime GitHub token is available, verify the remote hash through a temporary authenticated URL and unset it immediately; do not set token-bearing remotes or print the token. See `references/dashboard-usability-simplification.md` and `references/dashboard-user-mode-finalization-and-push.md`.\\n- Breaking an explicit MVP freeze by adding more providers/features after the user accepted the sprint. In freeze mode, produce the sanitized Zwischenbericht/Abnahmeplan, separate implemented vs not-finished vs risks, include concrete user acceptance checklists, and run the same verification/push discipline as code changes. See `references/mvp-freeze-zwischenbericht-abnahme.md`.\\n- During Budget/Cashflow runtime cleanup or UAT, allowing test/seed rows to leak into User Mode is a correctness bug. Create a runtime backup first, report dry-run counts only, archive/deactivate UAT artefacts with audit, then verify user-facing APIs/pages return zero matching visible objects. If you create temporary runtime UAT rows to prove Confirm/category/ignore flows, archive them immediately after verification. See `references/budget-phase18-core-confirm-flow.md`.\\n- Rebuilding SQLite tables with `ALTER TABLE ... RENAME TO ...` without checking dependent foreign keys. SQLite can rewrite dependent table references to the temporary table name, causing later inserts to fail with `no such table: main.`. After migrations that rebuild budget transactions/candidates, inspect `sqlite_master` for `__phase` references and regression-test dependent inserts into tags/transfers/line-items/splits. See `references/budget-phase18-core-confirm-flow.md`.\\n- Running Git-safety immediately after pytest without cleaning generated caches. Pytest can recreate `.pytest_cache`; remove `.pytest_cache` and `__pycache__` after the final test run, then run Git-safety before staging/commit/push.\\n- In MVP acceptance mode, exposing raw internal workflow states (`ok`, `missing`, `manual_override_required`, `not_needed`) to User Mode is a bug, not documentation. Convert them into plain choices such as automatic FX, manual FX, or save incomplete; keep the stored internal status deterministic and audited.\\n- Treating absent CoinGecko/Crypto detailcache code as something to invent during an acceptance bugfix sprint. If the codebase has no implementation, report it as an explicit open gap unless the user authorizes a new feature mini-sprint.\\n- Relying on the user's prose description of a spreadsheet layout instead of detecting the actual workbook structure. In adjusted crypto XLSX files, CoinGecko IDs may appear as an `API ID` column in the same row rather than a preceding row; use the detected structure, validate IDs, and report the difference aggregate-only.\\n- Treating formula/scientific-notation source cells as automatic import blockers. Count them, but block only when exact Decimal/Text conversion fails; never use float arithmetic for crypto quantities.\\n- Printing raw DOCX/table rows while analyzing real broker files. Use masked structural flags and aggregate counts only; real position names/values/quantities stay out of chat/log summaries.\\n- Counting ticker-like tokens from DOCX as confirmed Equity/ETF mappings. Without ISIN plus exchange/provider confirmation, mappings are `probable` or `needs_manual_review`, not productive-import ready.\\n- Treating low-confidence Equity/ETF DOCX extraction as enough for import. If most positions lack ISIN, cost basis or transaction history, build/operate a staging resolver and Mapping Review Queue first. Ticker-only rows stay `needs_manual_review`; local Qwen may help extract candidates but must not decide mappings or write ledger rows. See `references/equity-docx-instrument-resolver-review-workflow.md`.\\n- Treating verifier/file-mutation warnings as either automatically true or automatically false. Check the exact commit with `git show --name-status --patch -- `, then separately verify whether the test matrix actually covers the promised cases.\\n- Treating provider lookup smoke-test failures as simple `not reachable` / `no candidates` when runtime keys exist. Follow `references/provider-secret-auth-debugging.md`: confirm the runtime secret file is loaded without printing values, verify OpenFIGI uses `X-OPENFIGI-APIKEY` header and FMP uses `apikey`, split network reachability from provider auth, and classify HTTP 401/403 as `auth_failed`, 429 as `rate_limited`, empty 2xx as `no_results`, and DNS/timeout as `network_error`. Do not start a new feature block until the secret/auth diagnosis is resolved.\\n- Letting multi-provider market-data integration become implicit dashboard behavior or an API-call spray. Follow `references/provider-integration-sprint-v2.md`: normal dashboard render must use local DB/non-probing status only; provider calls require explicit click/CLI/schedule; FMP remains primary for search/price preview; Twelve Data/Finnhub/Massive are fallbacks; EODHD is never default and requires explicit low-volume opt-in; provider statuses and errors must be sanitized categories, not raw responses or secret-bearing URLs.\\n- Adding a browser frontend without expanding Git-safety and secret boundaries. For FinanceManager Vue/FastAPI work, block `frontend/dist`, `frontend/node_modules`, `.vite`, and `.env.*` except `.env.example`; provider keys stay backend/runtime-only and must not appear in `VITE_*` variables. FastAPI tests that inject an in-memory SQLite connection through `TestClient` need `check_same_thread=False`, otherwise SQLite may fail in the worker thread.\\n- Implementing Vue auto-load as hidden provider refresh. Auto-load is allowed only for local FastAPI/runtime DTOs; render must not call CoinGecko/OpenFIGI/FMP/Finnhub/Yahoo/etc. Add visible local-data status, explicit refresh semantics, and browser performance checks for zero external provider resources. For launch/browser sanity/process hygiene, follow `references/vue-dashboard-autoload-runtime-operations.md`.\\n- Handing the user a remote/Tailscale Vue link while the frontend still calls `http://127.0.0.1:8000` from the browser. For FinanceManager dev servers, either use same-origin `/api` calls through Vite's proxy or start Vite with `VITE_API_BASE_URL=http://:5173`, then verify both `http://127.0.0.1:5173/api/health` and the user-visible `http://:5173/api/health`. Do not expose provider/API secrets through `VITE_*`.\\n- Joining crypto holdings directly to historical/current price rows in detail or wallet readmodels. Multiple price rows per asset can duplicate wallet allocations and inflate values. Aggregate holdings by asset/wallet first, fetch the latest price once per asset, then value the aggregate; regression-test one asset split across wallets plus multiple price rows. See `references/crypto-vue-dashboard-readmodel-hardening.md`.\\n- Shipping a Vue/FastAPI cockpit update after tests pass but before real browser/API sanity from the current checkout. Stale uvicorn/node processes can mask route changes; if OpenAPI or click behavior looks stale, check/kill the process bound to the port and restart from the working tree. Verify `/positions/add`, `/equity` row-click drawers, `/openapi.json` endpoint registration, frontend tests/build from `frontend/`, changed-file/build-artifact secret scans, commit, push, and remote hash equality. See `references/vue-cockpit-v11-preview-confirm-drawers.md`.\\n- During Vue/FastAPI UAT, letting `JARVIS_FINANCE_DB_PATH` leak from the UAT server environment into pytest can make unit tests hit the UAT DB instead of their temporary runtime; unset it before full pytest unless the test explicitly requires a fixed DB path. For manual Cash and Equity/ETF Preview → Confirm flows, verify both persisted domain state and audit rows; Cash may currently need a bridge write to both ledger `transactions` and `cash_balances` compatibility/control rows. For the v1.2 completion pass, use `references/vue-manual-entry-v12-acceptance-completion.md`: update frontend mocks when moving from bare `searchInstruments` to provider-search DTOs, archive review/test accounts even when they have no remaining equity transactions, remove `.pytest_cache` and `frontend/dist` after tests/build before Git-safety, and verify authenticated push by remote-hash equality without printing or storing the token. For provider-search/preview-confirm hardening, use `references/vue-provider-preview-confirm-hardening.md`: Preview must not materialize provider/catalog instruments, Confirm is the write boundary, provider dedupe must preserve ISIN cross-listings by exchange/currency, cached CoinGecko URLs must be sanitized before DTO/DOM use, and independent re-review should focus on these exact failure modes. See `references/vue-cockpit-uat-preview-confirm-acceptance.md`.\\n- Treating external Vue/finance-dashboard references as a license-free component source or as authorization for a new feature sprint. When the user provides reference apps, first produce a reference analysis, design matrix, small UI design system, layout sketch, and component hierarchy; then apply only controlled read-only UX improvements. Copy no code/assets/branding, keep Streamlit as Admin/Fallback if applicable, avoid normal-render API calls, and park budgeting/FIRE/forecast/goal modules as roadmap unless explicitly authorized. See `references/vue-ux-reference-sprint-pattern.md`.\\n\", \"path\": \"finance/finance-system-architecture/SKILL.md\", \"skill_dir\": \"/home/agent/.hermes/skills/finance/finance-system-architecture\", \"linked_files\": {\"references\": [\"references/crypto-single-coin-runtime-test.md\", \"references/guided-manual-review-pilot.md\", \"references/multi-wallet-crypto-and-equity-manage-ui.md\", \"references/mvp-crypto-report-export.md\", \"references/broker-review-queue-hardening.md\", \"references/mvp-import-pipeline-tasks-9-13.md\", \"references/broker-mini-import-completion-and-dq-consolidation.md\", \"references/broker-parser-dry-run-pipeline.md\", \"references/crypto-xlsx-initial-snapshot-dry-run.md\", \"references/crypto-price-refresh-and-idempotent-btc-runtime.md\", \"references/provider-secret-auth-debugging.md\", \"references/spec-v0.3-decisions.md\", \"references/runtime-backup-and-broker-mapping.md\", \"references/mvp-dashboard-workflow-regressions.md\", \"references/vue-manual-entry-v12-acceptance-completion.md\", \"references/controlled-real-broker-dry-run.md\", \"references/vue-ux-reference-sprint-pattern.md\", \"references/broker-mini-import-regression-and-corrections.md\", \"references/equity-fx-market-data-phase-c.md\", \"references/true-wealth-valuation-pilot-c2.md\", \"references/mvp-market-import-safety-hardening.md\", \"references/budget-module-prd-drive-analysis.md\", \"references/spec-v0.4-implementation-blueprint.md\", \"references/budget-phase1-manual-ledger-foundation.md\", \"references/mvp-acceptance-dashboard-fx-valuation-cleanup.md\", \"references/vue-dashboard-autoload-runtime-operations.md\", \"references/equity-etf-provider-integration-openfigi-fmp.md\", \"references/mvp-freeze-zwischenbericht-abnahme.md\", \"references/mapping-candidate-ranking-decision-pack.md\", \"references/import-wizard-mapping-layer.md\", \"references/crypto-runtime-db-consistency-verification.md\", \"references/github-token-push-from-drive.md\", \"references/instrument-catalog-manual-add-workflow.md\", \"references/broker-import-execution-plan-mini-import.md\", \"references/mvp-streamlit-dashboard.md\", \"references/vue-manual-position-v122-hardening.md\", \"references/vue-dashboard-tailscale-api-base.md\", \"references/budget-phase18-core-confirm-flow.md\", \"references/spec-v0.4.1-corrections.md\", \"references/dashboard-real-ui-acceptance-regression.md\", \"references/mvp-crypto-wallet-ledger.md\", \"references/true-wealth-mapping-candidate-review-c3.md\", \"references/mvp-safety-gate-implementation.md\", \"references/budget-phase16-user-category-rules.md\", \"references/provider-integration-sprint-v2.md\", \"references/public-instrument-lookup-candidate-generator.md\", \"references/crypto-xlsx-initial-snapshot-production-import.md\", \"references/dashboard-usability-simplification.md\", \"references/spec-v0.5-repo-blueprint.md\", \"references/dashboard-user-mode-finalization-and-push.md\", \"references/spec-v0.2-decisions.md\", \"references/mvp-crypto-manual-management-ui.md\", \"references/vue-cockpit-uat-preview-confirm-acceptance.md\", \"references/fastapi-vue-dashboard-transition.md\", \"references/mvp-ledger-cash-wac-positions.md\", \"references/crypto-vue-dashboard-readmodel-hardening.md\", \"references/dashboard-functional-ux-stabilization.md\", \"references/equity-docx-instrument-resolver-review-workflow.md\", \"references/vue-provider-preview-confirm-hardening.md\", \"references/manual-position-entry-provider-ux.md\", \"references/budget-phase17-ux-workflow-refactor.md\", \"references/vue-cockpit-v11-preview-confirm-drawers.md\", \"references/budget-phase12-excel-seed-review.md\", \"references/crypto-adjusted-xlsx-and-equity-docx-dryrun.md\", \"references/dashboard-user-mode-ux-and-runtime-review.md\", \"references/budget-phase11-runtime-category-seed.md\", \"references/manual-equity-etf-entry-wizard.md\", \"references/git-safety-verification-coverage.md\", \"references/budget-source-aware-import-review.md\"]}, \"usage_hint\": \"To view linked files, call skill_view(name, file_path) where file_path is e.g. 'references/api.md' or 'assets/config.yaml'\", \"required_environment_variables\": [], \"required_commands\": [], \"missing_required_environment_variables\": [], \"missing_credential_files\": [], \"missing_required_commands\": [], \"setup_needed\": false, \"setup_skipped\": false, \"readiness_status\": \"available\", \"metadata\": {\"hermes\": {\"tags\": [\"finance\", \"portfolio\", \"accounting\", \"dashboard\", \"csv\", \"fx\", \"tax\", \"privacy\"]}}}",
"tool_call_id": "call_9y7KhTdFkxf5HfoGvIzkH05W"
},
{
"role": "assistant",
"content": "",
"reasoning": null,
"finish_reason": "tool_calls",
"tool_calls": [
{
"id": "call_LvdQXF7ubmPehK9k8haqNKrV",
"call_id": "call_LvdQXF7ubmPehK9k8haqNKrV",
"response_item_id": "fc_07795746cfedda55016a09c88716748191a36b5f3a911ae52f",
"type": "function",
"function": {
"name": "skill_view",
"arguments": "{\"name\":\"finance-system-architecture\",\"file_path\":\"references/budget-phase18-core-confirm-flow.md\"}"
}
},
{
"id": "call_IZ6YXCJuUGdyXxQaTq4yFTl4",
"call_id": "call_IZ6YXCJuUGdyXxQaTq4yFTl4",
"response_item_id": "fc_07795746cfedda55016a09c88716908191ace4eed11da7c073",
"type": "function",
"function": {
"name": "skill_view",
"arguments": "{\"name\":\"finance-system-architecture\",\"file_path\":\"references/budget-phase17-ux-workflow-refactor.md\"}"
}
},
{
"id": "call_AcTTbivVs1J6JsayqylVUmHg",
"call_id": "call_AcTTbivVs1J6JsayqylVUmHg",
"response_item_id": "fc_07795746cfedda55016a09c887169c8191a7f62fb3a3148b32",
"type": "function",
"function": {
"name": "skill_view",
"arguments": "{\"name\":\"github-pr-workflow\"}"
}
}
]
},
{
"role": "tool",
"name": "skill_view",
"content": "{\"success\": true, \"name\": \"finance-system-architecture\", \"file\": \"references/budget-phase18-core-confirm-flow.md\", \"content\": \"# Budget Phase 1.8 — Core candidate confirm flow and effective expenses\\n\\nUse this reference when Budget/Cashflow review pages or candidate-confirm behaviour are incorrect, duplicated, or show test/seed data as productive expenses.\\n\\n## Goal\\n\\nThe only acceptable core workflow is:\\n\\n`budget_transaction_candidate -> category set/overridden -> Confirm -> productive budget_transactions row -> Effective Expenses -> Budgetstatus`\\n\\nDo not add new import sources or broaden the feature surface while this flow is unstable.\\n\\n## Navigation / UX pattern\\n\\n- Avoid duplicating Import Review and Category Assignment as separate surfaces for the same rows.\\n- User Mode Budget menu should have a single central candidate work page, e.g. **Ausgaben -> Buchungen prüfen**.\\n- The central page owns: pending/import candidates, category proposal display, category override, Confirm, Ignore, Reopen, Split preparation, Transfer marking, details drawer and bulk category/confirm/ignore/reopen.\\n- Old routes may redirect to the central page to preserve bookmarks, but must not be visible as duplicate menu items.\\n- Confirm controls must be hidden/disabled unless the required fields are present, especially category and account.\\n\\n## Backend confirm contract\\n\\n`confirm_transaction_candidate` must validate before writing:\\n\\n- active category for expense/income candidates\\n- active account\\n- amount\\n- currency\\n- date\\n- description\\n\\nOn success it must:\\n\\n1. Insert a real `budget_transactions` row.\\n2. Set `status='confirmed'`.\\n3. Set `source_type='import_candidate'`.\\n4. Store `source_candidate_id` on `budget_transactions`.\\n5. Store `confirmed_transaction_id`, `confirmed_at`, `confirmed_by` on `budget_transaction_candidates`.\\n6. Write separate audit events for candidate confirmation and transaction creation, e.g. `candidate_confirmed` and `budget_transaction_created`.\\n7. Commit only after both productive row and candidate status are updated.\\n\\nDefault candidate list views should hide confirmed/ignored/non-booking statuses. Status-specific tabs or filters may still show `covered_by_migros`, `ignored`, etc. for review/admin needs.\\n\\n## Effective Expenses contract\\n\\nThe Effective Expenses page/API must show only productive confirmed expense transactions:\\n\\n- `budget_transactions.status='confirmed'`\\n- `transaction_type='expense'`\\n\\nIt must not show:\\n\\n- import candidates\\n- seed candidates\\n- ignored candidates\\n- `covered_by_migros`\\n- superseded article rows\\n- archived/test/UAT rows\\n\\nRow click/detail drawer may allow category/tag changes, but amount changes require adjustment/reversal flow. Category/tag changes must write audit and refresh Budgetstatus.\\n\\n## Budgetstatus coupling\\n\\nBudgetstatus by category must include categories that have confirmed actuals even if no budget plan exists yet. Otherwise a newly confirmed expense can appear under Effective Expenses but not in category status, which breaks acceptance.\\n\\nRow-click/category detail should query confirmed transactions by `category_id` rather than candidates.\\n\\n## Runtime cleanup pattern for UAT/test data\\n\\nBefore mutating runtime:\\n\\n1. Create a verified runtime DB backup outside Git using SQLite online backup where possible.\\n2. Dry-run cleanup and report aggregate counts only; never report real amounts.\\n3. Archive/deactivate test artefacts rather than hard-deleting unless explicitly required:\\n - budget accounts: `is_active=0`, `archived_at`\\n - budget categories/tags: `is_active=0`\\n - budget transactions: `status='archived'`\\n - candidates: `status='ignored'` or equivalent non-visible state\\n4. Write a cleanup audit event.\\n5. Verify user-facing views show zero matching test/UAT artefacts.\\n\\nFor temporary Runtime-UAT objects created by the agent, archive them again after verifying the flow, so Effective Expenses remains free of test data.\\n\\n## SQLite migration pitfall\\n\\nWhen rebuilding SQLite tables with `ALTER TABLE ... RENAME TO ...`, SQLite may rewrite foreign-key references in dependent tables to the temporary table name (e.g. `budget_transactions__phase18` or `budget_transaction_candidates__phase18`). This can later break inserts into related tables with `no such table: main.`.\\n\\nAfter table rebuilds, inspect `sqlite_master.sql` for `__phase` references and rebuild dependent tables or avoid FK references for cross-phase helper links where necessary. Tests should exercise inserts into dependent tables such as:\\n\\n- `budget_transaction_tags`\\n- `budget_transfers`\\n- `budget_import_line_items`\\n- `budget_candidate_splits`\\n\\n## Tests to add\\n\\nMinimum regression tests:\\n\\n- Confirm candidate creates `budget_transactions` row and stores `confirmed_transaction_id`.\\n- Confirmed candidate disappears from default pending/open candidate view.\\n- Effective Expenses lists the resulting confirmed expense transaction.\\n- Pending candidate category update is audited but does not book.\\n- Confirmed transaction category update is audited and updates Budgetstatus.\\n- Ignore hides candidate; Reopen makes it visible again.\\n- UAT/test artefacts are archived and absent from user-facing views.\\n- Navigation exposes **Buchungen prüfen** once and does not expose duplicate Import Review / Kategoriezuordnung menu entries.\\n- Migration has no lingering `__phase*` temporary table references in `sqlite_master`.\\n\\n## Verification\\n\\nRun in this order:\\n\\n1. Focused backend tests for the core flow.\\n2. Full `pytest` with `JARVIS_FINANCE_DB_PATH` unset unless explicitly testing runtime.\\n3. Frontend vitest.\\n4. Frontend build.\\n5. Apply migration to runtime and verify no temp FK references remain.\\n6. Changed-file secret scan and Git-safety; remove `frontend/dist`, `.vite`, `.pytest_cache`, `__pycache__` after tests/build.\\n7. Commit, push, and verify remote hash with authenticated temporary askpass if needed.\\n\", \"file_type\": \".md\"}",
"tool_call_id": "call_LvdQXF7ubmPehK9k8haqNKrV"
},
{
"role": "tool",
"name": "skill_view",
"content": "{\"success\": true, \"name\": \"finance-system-architecture\", \"file\": \"references/budget-phase17-ux-workflow-refactor.md\", \"content\": \"# Budget Phase 1.7 — UX-/Workflow-Refactor für Review, Ausgaben und Kategorien\\n\\nUse this reference when the user complains that the Budget dashboard is confusing, import review and category assignment are blurred, or visible actions are unreliable/fake. This is a UX/workflow stabilization class, not an import-source or analytics expansion.\\n\\n## Scope guard\\n\\n- Do not add new import sources.\\n- Do not add large analytics features.\\n- Do not commit runtime DBs, real CSV/XLS/XLSX/PDF/DOCX, reports, or build artifacts.\\n- Do not print real financial values in chat; local dashboard display is allowed.\\n- Do not create silent mutations. Writes must cross Preview/Confirm or explicit user action and write Audit.\\n- Import candidates remain candidates until Confirm creates a budget transaction.\\n- Confirmed expenses are real `budget_transactions` and must live in a separate effective-expenses workflow.\\n- Category changes, candidate status changes, transaction edits, and merges must be audit-visible.\\n- Undo is via reopen/reverse/status transition/audit trail, not destructive hidden overwrite.\\n\\n## Recommended workflow shape\\n\\n1. Re-check repo state before editing: branch, HEAD, `git status --short`, and changed files.\\n2. Inventory existing Budget API/services/pages before patching.\\n3. Separate the navigation into clear groups:\\n - Budget: overview, status, monthly/yearly/plans.\\n - Ausgaben: effective expenses, import review, manual expense, category assignment.\\n - Einkommen: actual/planned/forecast.\\n - Setup: categories, tags, accounts, seed review.\\n - Später: rules, fixed costs, duplicates, provider/detail pages not ready yet.\\n4. Add a short “Wofür ist diese Seite?” explanation to each work page.\\n5. Replace dominant read-only badges on work pages with explicit Review-Modus/Arbeitsmodus wording.\\n6. Make visible actions real. If a button cannot work now, hide it or mark it disabled with a reason.\\n\\n## Import Review pattern\\n\\nThe Import Review page is only for `budget_transaction_candidates`. It should show at least:\\n\\n- Datum\\n- Quelle / source file label / row or range\\n- Händler / Beschreibung\\n- Betrag and Währung in the local UI\\n- category suggestion as an editable dropdown\\n- confidence\\n- rule/classification/review reason\\n- status\\n- actions\\n\\nExpected actions:\\n\\n- Confirm creates a real `budget_transaction` and audits the candidate.\\n- Ignore marks candidate ignored and audits.\\n- Reopen moves ignored/review candidates back to a reviewable state and audits.\\n- Category dropdown updates `proposed_category_id` and audits.\\n- Bulk category assignment is allowed for selected candidates and audits.\\n- Safe auto candidates need a Preview endpoint and a later explicit Confirm; no blind batch booking.\\n- Transfer and split flows may be prepared/marked, but must not silently book expense/income.\\n\\n## Effective expenses pattern\\n\\nConfirmed expenses need a separate page over `budget_transactions`, not a mixed import-review list. Recommended controls:\\n\\n- Filter by month, account, category, merchant/source/search/tag/status.\\n- Table shows confirmed transaction rows with local amounts.\\n- Row click opens a detail drawer.\\n- Category/description/tag changes call audited update endpoints.\\n- Amount overwrites should be rejected; use adjustment/reversal flows instead.\\n\\n## Category assignment pattern\\n\\nCreate a dedicated page for categorizing:\\n\\n- review-required/open candidates (updates candidate proposal only), and\\n- confirmed uncategorized transactions (updates transaction category).\\n\\nSupport bulk assignment and quick category creation, with audit. Keep it distinct from both Import Review and Category Tree management.\\n\\n## Category tree + merge pattern\\n\\nThe Categories page should manage the tree:\\n\\n- Create category/subcategory via Preview → Confirm.\\n- Rename/change parent/type through an audited update endpoint.\\n- Delete only truly empty categories.\\n- Archive non-empty categories.\\n- Show transaction count / active budget plan count / active status.\\n- Category merge must be Preview → Confirm: move transactions, budget plans and children to target, archive sources, and audit the merge.\\n\\n## Backend service patterns\\n\\n- Keep `list_budget_transactions` filters table-qualified when joins exist (`t.category_id`, `t.status`, etc.).\\n- Support `category_id='__uncategorized'` as an explicit filter for confirmed but uncategorized transactions.\\n- Use `record_audit_event` for category updates/deletes/archives/merges, candidate status/proposal changes, and transaction edits.\\n- Idempotent category creation can return the existing category, but still audit as idempotent confirm.\\n\\n## Tests and verification\\n\\nAdd focused unit tests for:\\n\\n- Import Review: bulk category, ignore, reopen, safe preview, confirm creates transaction only at confirm boundary.\\n- Effective expenses: filters, category update with audit, amount overwrite rejected.\\n- Category tree: delete empty, archive/merge non-empty, moved transactions/plans after merge.\\n- Frontend: navigation copy, work-mode wording, table/drawer/category controls.\\n\\nRun and report:\\n\\n- `unset JARVIS_FINANCE_DB_PATH`\\n- `python -m compileall src tests`\\n- `pytest -q`\\n- from `frontend/`: `npm test -- --run`\\n- from `frontend/`: `npm run build`\\n- `git diff --check`\\n- source and build-artifact secret scans\\n- Git-safety scan for runtime/build/data artifacts\\n- cleanup: `frontend/dist`, `.pytest_cache`, `__pycache__`, `.vite` before final Git-safety/commit\\n- commit, push, and verify local `HEAD` equals remote branch hash via authenticated `git ls-remote` when needed.\\n\\n## Pitfalls learned\\n\\n- Do not let tests depend on wall-clock freshness with fixed timestamps. If a fixture timestamp crosses the stale threshold, update the fixture or make the test deterministic.\\n- Vue unit tests that mount route-aware components without router injection may warn; either provide a router or make route reads null-safe.\\n- After `npm run build`, scan `frontend/dist` before deleting it; then delete it before Git-safety and commit.\\n- Do not keep old test assertions for removed/relocated workflows such as tags on the Categories page when Phase 1.7 deliberately separates setup concerns.\\n\", \"file_type\": \".md\"}",
"tool_call_id": "call_IZ6YXCJuUGdyXxQaTq4yFTl4"
},
{
"role": "tool",
"name": "skill_view",
"content": "{\"success\": true, \"name\": \"github-pr-workflow\", \"description\": \"GitHub PR lifecycle: branch, commit, open, CI, merge.\", \"tags\": [\"GitHub\", \"Pull-Requests\", \"CI/CD\", \"Git\", \"Automation\", \"Merge\"], \"related_skills\": [\"github-auth\", \"github-code-review\"], \"content\": \"---\\nname: github-pr-workflow\\ndescription: \\\"GitHub PR lifecycle: branch, commit, open, CI, merge.\\\"\\nversion: 1.1.0\\nauthor: Hermes Agent\\nlicense: MIT\\nplatforms: [linux, macos, windows]\\nmetadata:\\n hermes:\\n tags: [GitHub, Pull-Requests, CI/CD, Git, Automation, Merge]\\n related_skills: [github-auth, github-code-review]\\n---\\n\\n# GitHub Pull Request Workflow\\n\\nComplete guide for managing the PR lifecycle. Each section shows the `gh` way first, then the `git` + `curl` fallback for machines without `gh`.\\n\\nFor private HTTPS remotes where an agent must push with a token but PR creation may be blocked by token scope, see `references/private-repo-token-push-pr.md`.\\n\\n## Prerequisites\\n\\n- Authenticated with GitHub (see `github-auth` skill)\\n- Inside a git repository with a GitHub remote\\n\\n### Quick Auth Detection\\n\\n```bash\\n# Determine which method to use throughout this workflow\\nif command -v gh &>/dev/null && gh auth status &>/dev/null; then\\n AUTH=\\\"gh\\\"\\nelse\\n AUTH=\\\"git\\\"\\n # Ensure we have a token for API calls\\n if [ -z \\\"$GITHUB_TOKEN\\\" ]; then\\n if [ -f ~/.hermes/.env ] && grep -q \\\"^GITHUB_TOKEN=\\\" ~/.hermes/.env; then\\n GITHUB_TOKEN=$(grep \\\"^GITHUB_TOKEN=\\\" ~/.hermes/.env | head -1 | cut -d= -f2 | tr -d '\\\\n\\\\r')\\n elif grep -q \\\"github.com\\\" ~/.git-credentials 2>/dev/null; then\\n GITHUB_TOKEN=$(grep \\\"github.com\\\" ~/.git-credentials 2>/dev/null | head -1 | sed 's|https://[^:]*:\\\\([^@]*\\\\)@.*|\\\\1|')\\n fi\\n fi\\nfi\\necho \\\"Using: $AUTH\\\"\\n```\\n\\n### Extracting Owner/Repo from the Git Remote\\n\\nMany `curl` commands need `owner/repo`. Extract it from the git remote:\\n\\n```bash\\n# Works for both HTTPS and SSH remote URLs\\nREMOTE_URL=$(git remote get-url origin)\\nOWNER_REPO=$(echo \\\"$REMOTE_URL\\\" | sed -E 's|.*github\\\\.com[:/]||; s|\\\\.git$||')\\nOWNER=$(echo \\\"$OWNER_REPO\\\" | cut -d/ -f1)\\nREPO=$(echo \\\"$OWNER_REPO\\\" | cut -d/ -f2)\\necho \\\"Owner: $OWNER, Repo: $REPO\\\"\\n```\\n\\n---\\n\\n## 1. Branch Creation\\n\\nThis part is pure `git` — identical either way:\\n\\n```bash\\n# Make sure you're up to date\\ngit fetch origin\\ngit checkout main && git pull origin main\\n\\n# Create and switch to a new branch\\ngit checkout -b feat/add-user-authentication\\n```\\n\\nBranch naming conventions:\\n- `feat/description` — new features\\n- `fix/description` — bug fixes\\n- `refactor/description` — code restructuring\\n- `docs/description` — documentation\\n- `ci/description` — CI/CD changes\\n\\n## 2. Making Commits\\n\\nUse the agent's file tools (`write_file`, `patch`) to make changes, then commit:\\n\\n```bash\\n# Stage specific files\\ngit add src/auth.py src/models/user.py tests/test_auth.py\\n\\n# Commit with a conventional commit message\\ngit commit -m \\\"feat: add JWT-based user authentication\\n\\n- Add login/register endpoints\\n- Add User model with password hashing\\n- Add auth middleware for protected routes\\n- Add unit tests for auth flow\\\"\\n```\\n\\nCommit message format (Conventional Commits):\\n```\\ntype(scope): short description\\n\\nLonger explanation if needed. Wrap at 72 characters.\\n```\\n\\nTypes: `feat`, `fix`, `refactor`, `docs`, `test`, `ci`, `chore`, `perf`\\n\\n## 3. Pushing and Creating a PR\\n\\n### Push the Branch (same either way)\\n\\n```bash\\ngit push -u origin HEAD\\n```\\n\\nIf the remote is HTTPS and interactive username/password prompts fail in an agent/non-TTY environment, use a temporary `GIT_ASKPASS` helper instead of embedding tokens in the remote URL. Keep the token in an environment variable, never echo it, and remove the helper after push/verification:\\n\\n```bash\\n# Ensure the askpass child process can see the token.\\nexport GITHUB_TOKEN\\nASKPASS=$(mktemp)\\ncat > \\\"$ASKPASS\\\" <<'SH'\\n#!/bin/sh\\ncase \\\"$1\\\" in\\n *Username*) printf '%s\\\\n' x-access-token ;;\\n *Password*) printf '%s\\\\n' \\\"$GITHUB_TOKEN\\\" ;;\\n *) printf '\\\\n' ;;\\nesac\\nSH\\nchmod 700 \\\"$ASKPASS\\\"\\ntrap 'rm -f \\\"$ASKPASS\\\"' EXIT\\nGIT_ASKPASS=\\\"$ASKPASS\\\" GIT_TERMINAL_PROMPT=0 git push -u origin HEAD\\n\\n# Optional authenticated remote-hash verification for private repos.\\nBRANCH=$(git branch --show-current)\\nLOCAL=$(git rev-parse HEAD)\\nREMOTE=$(GIT_ASKPASS=\\\"$ASKPASS\\\" GIT_TERMINAL_PROMPT=0 git ls-remote origin \\\"refs/heads/$BRANCH\\\" | awk '{print $1}')\\ntest \\\"$LOCAL\\\" = \\\"$REMOTE\\\"\\n\\nrm -f \\\"$ASKPASS\\\"\\n```\\n\\nThe `export GITHUB_TOKEN` line is not cosmetic: `GIT_ASKPASS` executes as a child process, so a shell variable that is not exported will look empty and GitHub will reject the push as an invalid username/token even when token extraction succeeded.\\n\\nDo not set token-bearing remotes, print token values, or leave askpass scripts behind.\\n\\nWhen extracting a token from a local CSV/secret file for `GIT_ASKPASS`, select by token shape/prefix (`github_pat_` or `ghp_`) rather than by excluding strings that start with `github`. Fine-grained GitHub tokens intentionally start with `github_pat_`; skipping all `github*` values can silently pass an empty/wrong password and produce `Invalid username or token`. Inspect only row/column counts or token lengths/prefix classes, never the token value.\\n\\n**Important permission pitfall:** a token that can push via HTTPS is not guaranteed to have REST/GraphQL permission to create PRs. After push, verify local/remote hash equality first. If PR creation returns `403 Forbidden`, do not retry by printing or broadening token handling in chat; report the pushed branch and GitHub's `/pull/new/` URL so the user can create the PR manually or provide a token with the required repo/pull-request scope.\\n\\n### Create the PR\\n\\n**With gh:**\\n\\n```bash\\ngh pr create \\\\\\n --title \\\"feat: add JWT-based user authentication\\\" \\\\\\n --body \\\"## Summary\\n- Adds login and register API endpoints\\n- JWT token generation and validation\\n\\n## Test Plan\\n- [ ] Unit tests pass\\n\\nCloses #42\\\"\\n```\\n\\nOptions: `--draft`, `--reviewer user1,user2`, `--label \\\"enhancement\\\"`, `--base develop`\\n\\n**With git + curl:**\\n\\n```bash\\nBRANCH=$(git branch --show-current)\\n\\ncurl -s -X POST \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n -H \\\"Accept: application/vnd.github.v3+json\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/pulls \\\\\\n -d \\\"{\\n \\\\\\\"title\\\\\\\": \\\\\\\"feat: add JWT-based user authentication\\\\\\\",\\n \\\\\\\"body\\\\\\\": \\\\\\\"## Summary\\\\nAdds login and register API endpoints.\\\\n\\\\nCloses #42\\\\\\\",\\n \\\\\\\"head\\\\\\\": \\\\\\\"$BRANCH\\\\\\\",\\n \\\\\\\"base\\\\\\\": \\\\\\\"main\\\\\\\"\\n }\\\"\\n```\\n\\nThe response JSON includes the PR `number` — save it for later commands.\\n\\nTo create as a draft, add `\\\"draft\\\": true` to the JSON body.\\n\\n## 4. Monitoring CI Status\\n\\n### Check CI Status\\n\\n**With gh:**\\n\\n```bash\\n# One-shot check\\ngh pr checks\\n\\n# Watch until all checks finish (polls every 10s)\\ngh pr checks --watch\\n```\\n\\n**With git + curl:**\\n\\n```bash\\n# Get the latest commit SHA on the current branch\\nSHA=$(git rev-parse HEAD)\\n\\n# Query the combined status\\ncurl -s \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/commits/$SHA/status \\\\\\n | python3 -c \\\"\\nimport sys, json\\ndata = json.load(sys.stdin)\\nprint(f\\\\\\\"Overall: {data['state']}\\\\\\\")\\nfor s in data.get('statuses', []):\\n print(f\\\\\\\" {s['context']}: {s['state']} - {s.get('description', '')}\\\\\\\")\\\"\\n\\n# Also check GitHub Actions check runs (separate endpoint)\\ncurl -s \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/commits/$SHA/check-runs \\\\\\n | python3 -c \\\"\\nimport sys, json\\ndata = json.load(sys.stdin)\\nfor cr in data.get('check_runs', []):\\n print(f\\\\\\\" {cr['name']}: {cr['status']} / {cr['conclusion'] or 'pending'}\\\\\\\")\\\"\\n```\\n\\n### Poll Until Complete (git + curl)\\n\\n```bash\\n# Simple polling loop — check every 30 seconds, up to 10 minutes\\nSHA=$(git rev-parse HEAD)\\nfor i in $(seq 1 20); do\\n STATUS=$(curl -s \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/commits/$SHA/status \\\\\\n | python3 -c \\\"import sys,json; print(json.load(sys.stdin)['state'])\\\")\\n echo \\\"Check $i: $STATUS\\\"\\n if [ \\\"$STATUS\\\" = \\\"success\\\" ] || [ \\\"$STATUS\\\" = \\\"failure\\\" ] || [ \\\"$STATUS\\\" = \\\"error\\\" ]; then\\n break\\n fi\\n sleep 30\\ndone\\n```\\n\\n## 5. Auto-Fixing CI Failures\\n\\nWhen CI fails, diagnose and fix. This loop works with either auth method.\\n\\n### Step 1: Get Failure Details\\n\\n**With gh:**\\n\\n```bash\\n# List recent workflow runs on this branch\\ngh run list --branch $(git branch --show-current) --limit 5\\n\\n# View failed logs\\ngh run view --log-failed\\n```\\n\\n**With git + curl:**\\n\\n```bash\\nBRANCH=$(git branch --show-current)\\n\\n# List workflow runs on this branch\\ncurl -s \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n \\\"https://api.github.com/repos/$OWNER/$REPO/actions/runs?branch=$BRANCH&per_page=5\\\" \\\\\\n | python3 -c \\\"\\nimport sys, json\\nruns = json.load(sys.stdin)['workflow_runs']\\nfor r in runs:\\n print(f\\\\\\\"Run {r['id']}: {r['name']} - {r['conclusion'] or r['status']}\\\\\\\")\\\"\\n\\n# Get failed job logs (download as zip, extract, read)\\nRUN_ID=\\ncurl -s -L \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/actions/runs/$RUN_ID/logs \\\\\\n -o /tmp/ci-logs.zip\\ncd /tmp && unzip -o ci-logs.zip -d ci-logs && cat ci-logs/*.txt\\n```\\n\\n### Step 2: Fix and Push\\n\\nAfter identifying the issue, use file tools (`patch`, `write_file`) to fix it:\\n\\n```bash\\ngit add \\ngit commit -m \\\"fix: resolve CI failure in \\\"\\ngit push\\n```\\n\\n### Step 3: Verify\\n\\nRe-check CI status using the commands from Section 4 above.\\n\\n### Auto-Fix Loop Pattern\\n\\nWhen asked to auto-fix CI, follow this loop:\\n\\n1. Check CI status → identify failures\\n2. Read failure logs → understand the error\\n3. Use `read_file` + `patch`/`write_file` → fix the code\\n4. `git add . && git commit -m \\\"fix: ...\\\" && git push`\\n5. Wait for CI → re-check status\\n6. Repeat if still failing (up to 3 attempts, then ask the user)\\n\\n## 6. Merging\\n\\n**With gh:**\\n\\n```bash\\n# Squash merge + delete branch (cleanest for feature branches)\\ngh pr merge --squash --delete-branch\\n\\n# Enable auto-merge (merges when all checks pass)\\ngh pr merge --auto --squash --delete-branch\\n```\\n\\n**With git + curl:**\\n\\n```bash\\nPR_NUMBER=\\n\\n# Merge the PR via API (squash)\\ncurl -s -X PUT \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/pulls/$PR_NUMBER/merge \\\\\\n -d \\\"{\\n \\\\\\\"merge_method\\\\\\\": \\\\\\\"squash\\\\\\\",\\n \\\\\\\"commit_title\\\\\\\": \\\\\\\"feat: add user authentication (#$PR_NUMBER)\\\\\\\"\\n }\\\"\\n\\n# Delete the remote branch after merge\\nBRANCH=$(git branch --show-current)\\ngit push origin --delete $BRANCH\\n\\n# Switch back to main locally\\ngit checkout main && git pull origin main\\ngit branch -d $BRANCH\\n```\\n\\nMerge methods: `\\\"merge\\\"` (merge commit), `\\\"squash\\\"`, `\\\"rebase\\\"`\\n\\n### Enable Auto-Merge (curl)\\n\\n```bash\\n# Auto-merge requires the repo to have it enabled in settings.\\n# This uses the GraphQL API since REST doesn't support auto-merge.\\nPR_NODE_ID=$(curl -s \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/repos/$OWNER/$REPO/pulls/$PR_NUMBER \\\\\\n | python3 -c \\\"import sys,json; print(json.load(sys.stdin)['node_id'])\\\")\\n\\ncurl -s -X POST \\\\\\n -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\\\\n https://api.github.com/graphql \\\\\\n -d \\\"{\\\\\\\"query\\\\\\\": \\\\\\\"mutation { enablePullRequestAutoMerge(input: {pullRequestId: \\\\\\\\\\\\\\\"$PR_NODE_ID\\\\\\\\\\\\\\\", mergeMethod: SQUASH}) { clientMutationId } }\\\\\\\"}\\\"\\n```\\n\\n## 7. Complete Workflow Example\\n\\n```bash\\n# 1. Start from clean main\\ngit checkout main && git pull origin main\\n\\n# 2. Branch\\ngit checkout -b fix/login-redirect-bug\\n\\n# 3. (Agent makes code changes with file tools)\\n\\n# 4. Commit\\ngit add src/auth/login.py tests/test_login.py\\ngit commit -m \\\"fix: correct redirect URL after login\\n\\nPreserves the ?next= parameter instead of always redirecting to /dashboard.\\\"\\n\\n# 5. Push\\ngit push -u origin HEAD\\n\\n# 6. Create PR (picks gh or curl based on what's available)\\n# ... (see Section 3)\\n\\n# 7. Monitor CI (see Section 4)\\n\\n# 8. Merge when green (see Section 6)\\n```\\n\\n## Useful PR Commands Reference\\n\\n| Action | gh | git + curl |\\n|--------|-----|-----------|\\n| List my PRs | `gh pr list --author @me` | `curl -s -H \\\"Authorization: token $GITHUB_TOKEN\\\" \\\"https://api.github.com/repos/$OWNER/$REPO/pulls?state=open\\\"` |\\n| View PR diff | `gh pr diff` | `git diff main...HEAD` (local) or `curl -H \\\"Accept: application/vnd.github.diff\\\" ...` |\\n| Add comment | `gh pr comment N --body \\\"...\\\"` | `curl -X POST .../issues/N/comments -d '{\\\"body\\\":\\\"...\\\"}'` |\\n| Request review | `gh pr edit N --add-reviewer user` | `curl -X POST .../pulls/N/requested_reviewers -d '{\\\"reviewers\\\":[\\\"user\\\"]}'` |\\n| Close PR | `gh pr close N` | `curl -X PATCH .../pulls/N -d '{\\\"state\\\":\\\"closed\\\"}'` |\\n| Check out someone's PR | `gh pr checkout N` | `git fetch origin pull/N/head:pr-N && git checkout pr-N` |\\n\", \"path\": \"github/github-pr-workflow/SKILL.md\", \"skill_dir\": \"/home/agent/.hermes/skills/github/github-pr-workflow\", \"linked_files\": {\"references\": [\"references/private-repo-token-push-pr.md\", \"references/conventional-commits.md\", \"references/ci-troubleshooting.md\"], \"templates\": [\"templates/pr-body-bugfix.md\", \"templates/pr-body-feature.md\"]}, \"usage_hint\": \"To view linked files, call skill_view(name, file_path) where file_path is e.g. 'references/api.md' or 'assets/config.yaml'\", \"required_environment_variables\": [], \"required_commands\": [], \"missing_required_environment_variables\": [], \"missing_credential_files\": [], \"missing_required_commands\": [], \"setup_needed\": false, \"setup_skipped\": false, \"readiness_status\": \"available\", \"metadata\": {\"hermes\": {\"tags\": [\"GitHub\", \"Pull-Requests\", \"CI/CD\", \"Git\", \"Automation\", \"Merge\"], \"related_skills\": [\"github-auth\", \"github-code-review\"]}}}",
"tool_call_id": "call_AcTTbivVs1J6JsayqylVUmHg"
},
{
"role": "assistant",
"content": "",
"reasoning": null,
"finish_reason": "tool_calls",
"codex_reasoning_items": [
{
"type": "reasoning",
"encrypted_content": "gAAAAABqCciJI7_u-f0EdT5lePcxy5YIJ0VwmkEd5a_mxtYt6_Xt7X7acMaJoKAOjv65XW-D5k5aDUUDqnGwAGCAgwrOOHTATt05-A1RdeilfA3Xha1ja2N4iMtxh0nKiDt8H20rpgUDamqPNQxxIsJeJ9P4gfz5dI9y3nykZD1OVobcxvnzKymqwXbtbSyTNxIO5PmmGgG4DO_eei-bUE38FHkNmBXWC9fWwJfyuZHbaAl-F_ATHOMstIleyiQH0q3Oc49zTDoKkT_LcDoKPdGQJyHDL43D0Yz5D7CryxDLzU2fBCSJrior3qDTF_ybZsvgtBgFkf90WszktJwpc0y8SIA1gg3MRucI93ImkPkfcCu8GeYS7IF-yliI2AwSSYcYNkLdEpbrhmBxH_3F4hQv7MJtB0yz9Aa-JKy7GG1JG-9AllTvrh_vByvyjk4CuWDptccCDRP4DCXgVTvX-HfRXJ-6dQHFPqkHDZJ1A3_bSkVXvnSPpvyJjJu4xaz266NkOUJPccWrKob0dVQ-EwXY6gs-PnX46cduK4o4HTlEjpT6KEeWtRQvEXzffwjMpvX2XHFtDDZHSf6ReuAJnuwdHUCU9upwy3Uke3jO67ebzZ7cxpDWq4dcPL3ogivRtOCwMcbFxR3QmUOLtZrPYYBNCQ7TJ2OfB6IQ-QUzaSoXWVmItitMLE8eOkF3YjOqwNQRDTuGWU8CRh2-nzzn71ORILdWnnyEoR4m1PCy2-hpAuT_oJ1QBaGcXIhyPG36XA_-VKs3qqUfXJJLq7B_fqrgT_vPJd62NMjghMWs9wCfOi3E7cgpKxFLLnCFAlyBAKUkiURxUOf67yuJJeYt5GyccNqMojE84cZdmXRGHzAVNNDqW1PRvnwExYyVgZTJBbgwOCPhzcjc6PgEoKstdeb3FHWDETYWCFXHKxAO2eZy0WaI1VaKA453pRuNjl--76Fnved-4LZE",
"id": "rs_0906abb75148f282016a09c888b4d88191b23638207408a1ae",
"summary": []
}
],
"tool_calls": [
{
"id": "call_SFZwQkmSJW0mJ3dtyB7z0Qrj",
"call_id": "call_SFZwQkmSJW0mJ3dtyB7z0Qrj",
"response_item_id": "fc_0906abb75148f282016a09c88941688191b76762266efd0cd4",
"type": "function",
"function": {
"name": "todo",
"arguments": "{\"todos\":[{\"id\":\"discover\",\"content\":\"Repo-/Runtime-Zustand prüfen und Sicherheitsgrenzen verifizieren\",\"status\":\"in_progress\"},{\"id\":\"backend\",\"content\":\"Backend-APIs/Services/Migrationen für Review, Batch, Split, Kategorien, Regeln, Ausgaben/Budget/Income erweitern\",\"status\":\"pending\"},{\"id\":\"frontend\",\"content\":\"Vue UI für Review Productivity, Split, Rule Manager, Ausgaben/Budget/Income ausbauen\",\"status\":\"pending\"},{\"id\":\"tests\",\"content\":\"Backend-/Frontend-Tests ergänzen und ausführen\",\"status\":\"pending\"},{\"id\":\"verify\",\"content\":\"Secret-/Git-Safety, Diff-Check, Commit, Push und Remote-Verifikation\",\"status\":\"pending\"}]}"
}
}
]
},
{
"role": "tool",
"name": "todo",
"content": "{\"todos\": [{\"id\": \"discover\", \"content\": \"Repo-/Runtime-Zustand prüfen und Sicherheitsgrenzen verifizieren\", \"status\": \"in_progress\"}, {\"id\": \"backend\", \"content\": \"Backend-APIs/Services/Migrationen für Review, Batch, Split, Kategorien, Regeln, Ausgaben/Budget/Income erweitern\", \"status\": \"pending\"}, {\"id\": \"frontend\", \"content\": \"Vue UI für Review Productivity, Split, Rule Manager, Ausgaben/Budget/Income ausbauen\", \"status\": \"pending\"}, {\"id\": \"tests\", \"content\": \"Backend-/Frontend-Tests ergänzen und ausführen\", \"status\": \"pending\"}, {\"id\": \"verify\", \"content\": \"Secret-/Git-Safety, Diff-Check, Commit, Push und Remote-Verifikation\", \"status\": \"pending\"}], \"summary\": {\"total\": 5, \"pending\": 4, \"in_progress\": 1, \"completed\": 0, \"cancelled\": 0}}",
"tool_call_id": "call_SFZwQkmSJW0mJ3dtyB7z0Qrj"
},
{
"role": "assistant",
"content": "",
"reasoning": null,
"finish_reason": "tool_calls",
"tool_calls": [
{
"id": "call_sSEp4otcdtEJCvydKwBmrnXW",
"call_id": "call_sSEp4otcdtEJCvydKwBmrnXW",
"response_item_id": "fc_04c3d96f9cdc9e63016a09c88f06888191b93c4e0c1ba0d7ee",
"type": "function",
"function": {
"name": "search_files",
"arguments": "{\"pattern\":\"FinanceManager\",\"path\":\"/home/agent\",\"target\":\"files\",\"limit\":20}"
}
}
]
},
{
"role": "tool",
"name": "search_files",
"content": "{\"total_count\": 1, \"files\": [\"[Command interrupted]\"]}",
"tool_call_id": "call_sSEp4otcdtEJCvydKwBmrnXW"
}
]
}