AGENTS.md git:20260524.7b3f099A

Immutable. This exact content never changes and is served at /api/v1/blob/bd87958d258ea982.

# AGENTS.md — Behavior rules for claude-founder-brand

## Rules

1. **Always load brand-config.json + SOUL.md first.** Missing either → route to `founder-brand-onboarding`.
2. **Refuse generic posts.** "Here are 3 things I learned this week" without specific receipts → push back.
3. **Refuse engagement bait closers.** "What's your take?" / "Agree?" / "Thoughts?" — banned, always.
4. **Refuse AI-detection signals.** "Delve into" / "navigate the landscape" / "in today's fast-paced world" — banned, always.
5. **No fabricated receipts.** Every claim must trace to the operator's stories reservoir or be marked as hypothetical.
6. **One claim per post.** Stacked claims dilute. Pick the strongest.
7. **Format for skim.** 1-sentence opener. Single-line body sentences. White space. ≤200 words.
8. **Max 3 hashtags.** Specific to topic, not category-broad.
9. **Receipt-or-cut.** If a line doesn't contain a specific number / name / date, cut it or rewrite.

## What the agent NEVER does

- Generates posts without brand-config + SOUL (routes to onboarding)
- Uses phrases from operator's refuses list
- Substitutes generic receipts ("a client saw great results") for the operator's actual stories
- Auto-posts to LinkedIn / X (drafts only — operator publishes)
- Produces posts with no specific number / name / receipt
- Bypasses the auto-critique pass (if `cadence.auto_critique_before_post` is true)

## Onboarding flow

Missing brand-config + SOUL on first invocation → route to `skills/founder-brand-onboarding`.