Skip to content

Add CLAUDE.md agent guide#1

Merged
RyanMorash merged 1 commit into
mainfrom
add-claude-md
May 21, 2026
Merged

Add CLAUDE.md agent guide#1
RyanMorash merged 1 commit into
mainfrom
add-claude-md

Conversation

@RyanMorash
Copy link
Copy Markdown
Member

@RyanMorash RyanMorash commented May 21, 2026

Summary

Adds a CLAUDE.md at the repo root to give Claude Code (and other agents) load-bearing context without requiring a full read of docs/architecture.md.

What's in it

  • Status note (pre-implementation) so agents don't hunt for code that isn't written yet.
  • uv run command list — non-obvious for tools trained on plain pytest/mypy.
  • Pointer to docs/architecture.md as authoritative for module layout and data contracts.
  • The architecture doc's most load-bearing "do not" rules surfaced inline:
    • No asyncio
    • Pure modules (reconciler, safety, tier_mapping) stay pure
    • Frozen dataclasses
    • Strict import layering
    • Card ID redaction in logs
    • Dry-run is sacred
    • Fail-secure on safety guards
  • Testing notes (pure-module tests use plain dataclasses; idempotency canary).
  • Config layout (env + TOML split).

Why

The architecture doc is thorough (430+ lines) but agents won't always read it before touching code. Surfacing the rules that, if violated, would silently corrupt the design intent — async migration, logging inside pure modules, full card IDs in logs — makes them bind without depending on the agent reading the full doc.

Kept short (~40 lines) on purpose: a longer file would duplicate docs/architecture.md and drift.

Test plan

  • Skim CLAUDE.md and confirm every rule still matches the current docs/architecture.md.
  • Verify the listed uv run commands work locally.

🤖 Generated with Claude Code

Surfaces the load-bearing invariants from docs/architecture.md (no
asyncio, pure modules stay pure, frozen dataclasses, card ID redaction,
fail-secure safety guards) so they bind without requiring a full read of
the architecture doc. Includes uv-prefixed command list and a pointer to
the architecture doc as authoritative.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings May 21, 2026 02:34
Copy link
Copy Markdown

Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Adds a root-level CLAUDE.md “agent guide” intended to surface the repo’s most important architecture constraints and common dev commands without requiring agents to read the full docs/architecture.md.

Changes:

  • Introduces CLAUDE.md with a concise project status note and workflow command list.
  • Summarizes “hard rules” from docs/architecture.md (sync-only, purity boundaries, layering, redaction, dry-run, safety fail-secure).
  • Documents testing and config layout expectations for future implementation work.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comment thread CLAUDE.md
Comment on lines +12 to +15
uv run mypy src tests # type check (strict)
uv run ruff check . # lint
uv run door-sync --once # one reconcile cycle, exit
uv run door-sync --dry-run # compute + log diff; no UniFi writes
Comment thread CLAUDE.md
Comment on lines +14 to +15
uv run door-sync --once # one reconcile cycle, exit
uv run door-sync --dry-run # compute + log diff; no UniFi writes
@RyanMorash RyanMorash merged commit 321e1c4 into main May 21, 2026
5 checks passed
@RyanMorash RyanMorash deleted the add-claude-md branch May 21, 2026 02:38
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants