# Everyday Red Flags — topic database and duration policy

Use this when selecting the next AutoShorts/TrueTraceShorts Proof Short after the May 2026 Everyday Red Flags expansion.

## Topic database location

The user supplied a large topic bank which has been normalized into JSON at:

- `/home/agent/jarvis_runtime/autoshortsbot/AutoShortsBot/data/content_ideas/everyday_red_flags_topic_database.json`
- `/home/agent/jarvis_runtime/AutoShortsBot/data/content_ideas/everyday_red_flags_topic_database.json`

Shape:

- `database_id`: `truetrace_everyday_red_flags_topic_database`
- 12 categories
- 185 topic items
- Each item has `topic_id`, `slug`, `title_de`, `series_line`, `status`, `priority`, `red_flag_prompt`, and `safe_action_prompt`.

When creating a new video, prefer topics with `priority = seed_next_batch` or normal backlog topics that fit the current channel mix. After rendering a candidate, mark the matching topic `status = review_ready`, set `existing_artifact_hint`, and change `priority = awaiting_review`. After upload, do not over-update with transient YouTube IDs; use audits/packages for those.

## Main category groups

1. SMS- und Messenger-Scams — `Texts that want you to hurry.`
2. E-Mail-Scams — `Emails that look normal.`
3. Banking- und Payment-Scams — `Money move red flags.`
4. Login- und Account-Fallen — `Account traps.`
5. Marketplace- und Kleinanzeigen-Scams — `Marketplace red flags.`
6. Fake Support / Tech-Support-Scams — `Real support does not hijack your screen.`
7. Social Media und Plattform-Fallen — `Creator account traps.`
8. Family- und Emotion-Scams — `Send this to your family.`
9. Job-, Karriere- und Nebenverdienst-Scams — `If they make you pay to work, stop.`
10. Shopping-, Abo- und Consumer-Traps — `Before you buy.`
11. Privacy- und App-Risiken — `App permission red flags.`
12. Technische Internet-Risiken, einfach erklärt — simple internet-safety basics.

## Already produced / avoid immediate duplicates

At the time this reference was written, these ideas had already been produced or uploaded/reviewed in the current run:

- `erf-001` delivery/package SMS
- `erf-002` Hi Mom / new number
- `erf-003` fake fraud / bank one-time-code
- `erf-004` parking QR payment sticker
- `erf-005` toll fee SMS
- `erf-006` account-lock phishing message
- `erf-007` fake support popup / call-support-now

Use the JSON status/audit files as source of truth if this list becomes stale.

## Duration policy

The user explicitly corrected that video length is not fixed. Choose length by viewer understanding:

- Simple red flag: 25–35s.
- Standard red flag: 35–45s.
- Explanation / recovery: 45–60s.
- TikTok monetization test / deeper lesson: 61–90s.
- Later longform guide: 5–8min.

Default for new Proof Shorts: 35–45s.

Priority order:

1. Viewer recognizes the scam/mechanism.
2. Viewer understands the red flag.
3. Viewer knows what to do next.
4. Video stays as short as possible, but not shorter than useful.

Early objective is shares, saves, followers, and trust — direct ad monetization is not the main initial target.

## Review / upload pattern reminder

For each new rendered Proof Short, deliver:

- Telegram-playable MP4 from Hermes media cache.
- Proposed title.
- Engaging YouTube description.
- QA block with duration, SAR/DAR, word/caption counts, video SHA256, posting pack SHA256, upload gate status.
- Exact hash-bound `APPROVED_FOR_PRIVATE_YOUTUBE_UPLOAD ...` command.
- Duration immediately after the approval code block.

If the user replies with natural language approval immediately after a single review package, reuse the saved package's stored approval command and run dry-run before upload. Do not require the user to paste the exact command unless multiple packages are in scope.