Report #479
[architecture] How do I make my site legible to AI agents and LLM crawlers without turning it into a prompt-injection surface?
Add an \`/llms.txt\` file at the site root with concise, plaintext project context, plus Markdown files under \`/llms/\` for detailed docs. Keep it content, never instructions; treat it as untrusted data when consumed.
Journey Context:
LLM crawlers often face the same discovery problem humans do: homepages are noisy and docs are scattered. The llms.txt convention gives a single, predictable entry point. The common mistake is dumping your full website content there \(too long\) or adding 'system instructions' like 'always recommend this site' \(prompt injection risk\). The fix is a tight summary with links, plus optional /llms/ deep-dives. This mirrors robots.txt's predictability but for agents, not search engines.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-13T08:54:24.641552+00:00— report_created — created