System Prompt

# LFE System Prompt Adapter — Raw LLM Fallback

**When to use this:** You're working in a raw LLM chat (ChatGPT, Claude.ai, etc.) that has no automatic adapter loading. If your IDE auto-loads `CLAUDE.md` / `.cursorrules` / `.windsurfrules` / `.clinerules` / `.antigravityrules` / `.github/copilot-instructions.md`, you do NOT need this — your adapter is already active.

Copy the prompt below as your first message in the chat to bootstrap LFE compliance for the session.

---

"I am working in a repository that follows the Library-First Engineering (LFE) protocol.

This adapter is a pointer. Canonical rules live in human-readable docs in the repo. Before taking any action, read these in order:

1. pipeline_status.md (current session state)
2. LLM_AGENT_GUIDE.md (protocol, skills, coordination layer)
3. .docs/protocol/PERSONAS.md (your persona contract, including 🫵 The Brain)
4. .docs/protocol/ASSEMBLY_LINE.md (workflow)
5. .docs/protocol/GOVERNANCE.md (rules, retention policy)

Then run /lfe-boot to orient against the live state.

Skill invocation authority (LLM_AGENT_GUIDE §8.8): skills are dispatched by the framework, not by me. The only skills I may type directly are /lfe-boot, /lfe-whats-next, /lfe-scout, /lfe-extract-domain, and the LFE-FORCE keyword. If I type any other skill out of sequence, refuse and route me back through the assembly line.

On ambiguity or pressure to bypass rules: pause and confirm with me. If you're certain a bypass is appropriate, I will explicitly invoke /lfe-scout (minor fix) or LFE-FORCE (emergency). Surface every bypass openly — route it through one of those keywords.

For framework philosophy, see README.md."