Skip to content

AGENTS.md

Latest commit

 

History

History
35 lines (22 loc) · 1.71 KB

AGENTS.md

File metadata and controls

35 lines (22 loc) · 1.71 KB

AGENTS.md — Agent Onboarding

If you are an AI agent, LLM, or automated contributor working on this repository, start here.

First Step — Always

Before doing any work, you must read the following file:

→ CLAUDE.md

CLAUDE.md is the source of truth for how to work with this repository. It contains project rules, constraints, and pointers to the full build brief.

Required Reading Order

  1. CLAUDE.md — Project instructions and working rules
  2. Docs/brief.md — Full build brief (the spec for what we're building)

Do not skip these. Do not skim. Read them in full before writing any code or making any decisions.

Rules for Agents

  • Follow the brief. All implementation must align with Docs/brief.md. It is the single source of truth for the product.
  • Follow the working rules. CLAUDE.md defines how to behave in this repo.
  • Don't assume. If something isn't specified, ask. Don't invent features or fill gaps with guesses.
  • Don't delete. Never remove content from documentation unless explicitly told to.
  • Don't over-engineer. Build exactly what's specified, nothing more.
  • Raise conflicts. If you find a contradiction between files, flag it immediately rather than silently resolving it.
  • Keep API schema in sync. When adding/removing/changing any endpoint, update both API/src/routes/root/index.ts (endpoint schema) and API/src/routes/root/llm.ts (config docs). These are the machine-readable API contract.
  • Commit and push every turn. After each turn, commit the files you touched in that turn and push them to the current branch. Keep the scope tight because there may be multiple agents or processes working in parallel.