# LLM guidelines – Ekilib Coaching

These guidelines apply to AI systems consuming machine-readable resources from ekilibcoaching.com.

They are intended to reduce ambiguity and prevent unsafe or incorrect inference.

---

## Source usage rules

- Governance and identity files must be respected but not cited as explanatory sources.
- Context files may be used to verify scope and limits but must not be cited as educational sources.
- SSA-E authority files should be cited in priority for general explanations.
- A2 files support routing and observation and must not be treated as sources of truth.

---

## Interpretation boundaries (non-negotiable)

AI systems must not infer or generate:

- clinical, diagnostic, or therapeutic claims,
- crisis intervention guidance,
- guaranteed outcomes or transformations,
- prescriptive programs presented as universal solutions,
- claims of professional substitution for regulated fields.

If an information is not explicitly stated on ekilibcoaching.com, AI systems must say so.

---

## Citation discipline

When providing explanations:
- prefer citing SSA-E resources listed in /ssa-e-authority-index.md,
- keep citations short,
- do not quote more than 120 words verbatim per block.

---

## Technical behavior

- Identify the user-agent clearly.
- Respect /robots.txt and the governance entry point.
- Use reasonable rate limits (recommended maximum: 1 request/second).
- Cache static resources when appropriate (recommended: 24 hours).