You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CLAUDE.md
-12Lines changed: 0 additions & 12 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -220,18 +220,6 @@ These are **non-negotiable** — never skip them:
220
220
221
221
6. **Commit messages**: Follow [Conventional Commits](https://www.conventionalcommits.org/en/v1.0.0/). Single line only, no description body. Examples: `docs: fix installation instructions for unsigned app`, `fix: prevent crash on empty query result`, `feat: add CSV export`.
222
222
223
-
## Agent Execution Strategy
224
-
225
-
- **Plans must include edge cases.** When creating implementation plans, identify edge cases, thread safety concerns, and boundary conditions. Include them as explicit checklist items in the plan — don't defer discovery to code review.
226
-
- **Implementation includes self-review.** Before committing, agents must check: thread safety (lock coverage, race conditions), all code paths (loops, early returns, between iterations), error handling, and flag/state reset logic. This eliminates the review→fix→review cycle.
227
-
- **Tests are part of implementation, not a separate step.** When implementing a feature, write tests in the same commit or immediately after — don't wait for a separate `/write-tests` invocation. The implementation agent should include test writing in its scope.
228
-
- **Always use team agents** for implementation work. Use the Agent tool (not subagents/tasks) to delegate coding to specialized agents (e.g., `feature-dev:feature-dev`, `feature-dev:code-architect`, `code-simplifier:code-simplifier`).
229
-
- **Always parallelize** independent tasks. Launch multiple agents in a single message.
230
-
- **Main context = orchestrator only.** Read files, launch agents, summarize results, update tracking. Never do heavy implementation directly.
231
-
- **Agent prompts must be self-contained.** Include file paths, the specific problem, and clear instructions.
232
-
- **Use worktree isolation** (`isolation: "worktree"`) for agents making code changes. This keeps the main branch clean and allows parallel work without conflicts.
233
-
- **Implementation standards** (apply to ALL new features and refactors): Clean architecture, correct macOS/Apple platform approach, proper design patterns, no backward compatibility hacks, easy to maintain and extensible. Always include these requirements in agent prompts.
234
-
235
223
## Performance Pitfalls
236
224
237
225
These have caused real production bugs — be aware when working in editor/autocomplete/persistence code:
0 commit comments