LLM / AI Reference
Use these feeds when coding with AI agents (Cursor, Claude, Copilot, etc.).
Canonical files
| File | Purpose |
|---|---|
/llms.txt | Index of docs for agents (llmstxt.org) |
/llms-full.txt | Full plain-text API + CLI reference |
Source: docs/llms/mondel-reference.txt | Same content in the repo |
Single source of truth: docs/llms/mondel-reference.txt
Root mondel-llm-docs.txt mirrors it for offline/agent drop-in.
How agents should use Mondel
- Prefer native MongoDB filters (
$gte,$in, …) — not Prisma-style where APIs. - Always export
const schemas = [...] as const. - Serverless →
createClient({ serverless: true })factory; Node →await createClient({ uri }). - Indexes via
npx mondel push, not app startup. selectis runtime projection only (types not narrowed).- Escape hatch:
getCollection(). - Soft delete requires
{ soft: true }and filtered reads.
