docs/`
Durable design documentation for Rome. [README.md](README.md) maps what lives where, and the per-content-type rulebooks live in [`authoring/`](authoring/authoring.md).
## Playbook
- Before writing any prose in `docs/`, read
Claude as a commit author.
- Always commit as using the default git settings
## Documentation Style
When creating or updating markdown documentation files:
- **Never create .md files unless explicitly instructed
SKILL.md` §3**: it must map to a real `DISPATCH` op or an `@tool`. A documented
command with no implementation is worse than no command.
- **Smoke-check:** `.venv/bin/python intel_engine/harness/mcp_server.py`, send
conversational replies (e.g., English prompt → English reply, 한국어 프롬프트 → 한국어 답변).
### Code & Documentation Language
- **English by default** for ALL generated content:
- Code, comments, commit messages, PR descriptions
- README.md, CHANGELOG.md
asks you to produce. Trigger words: *"write,"* *"save,"* *"summarize this into a doc,"* *"draft,"* *"document this,"* *"deep-research X and write it up."* The student wants something to keep
made reliable without fighting the framework, drop it and document the coverage gap in the PR body rather than papering over with `--repeat-each`.
Completion checklist: typecheck clean → new tests
ideas into production-ready code through specialized AI agents working in coordinated phases.
## Project Documentation Conventions (Important)
**Documentation Files:** All new documentation or task files must be saved under
Examples → Notes
- **`--raw` flag**: All commands support `--raw` for single-line JSON output. Always document it in the Notes section.
- **YAML frontmatter**: Quote `argument-hint` values that contain `|` characters
deployment-specific delta; the same thin-seam rules as downstream CMS branches apply.
## Sibling Documentation Repository
- `../xinshi-docs/stories/` is the single maintenance location for feature, architecture, API, operations, and WBS documentation
Sendable`.** Don't suppress data-race diagnostics
with `@unchecked Sendable` unless you have a documented reason.
- Use `actor` for stateful coordinators (e.g. `Conversation`,
`InMemorySession`), `struct` for value types, and `AsyncThrowingStream
verify they pass
- Write clear, focused commit messages (e.g., `Fix SMB2 negotiate response handling`)
#### Documentation updates when adding new tests
When adding new test cases, **all of the following must
claude/agents/skeptic.md`.
**ENGLISH ONLY**: keep all repository content in English, including source
comments, tests, technical documentation, user documentation and agent
configuration.
"When I report a bug, don't start by trying
/legacy`** - The stable, production-ready Simone system
- Directory-based command system
- Fully functional and documented
- Use this for actual project management
2. **`/hello-simone`** - NPM installer for legacy Simone
- Installs
errors to files, not console
3. **Testing** - Write tests for new features
4. **Documentation** - Update README and inline docs
5. **Performance** - Keep the server lightweight and fast
## Current Implementation Status
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
Where does it go?
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
What should it contain?
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
CLAUDE.md or AGENTS.md?
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.