bchanot/claude
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

docs(claude): make TODO.md systematic for complex write/modify tasks

Browse Source 
Replace the "3+ steps" threshold with a logic-based criterion: any
write/modify task touching new behavior, control flow, state, API,
or dependencies requires a plan in tasks/TODO.md decomposed into
subtasks — even a single complex task.

Drop the blanket "Confirm before implementing" step: confirmation is
now limited to real trade-offs (multiple valid approaches, breaking
changes, destructive actions) to avoid rubber-stamp friction.

Add an explicit exemption list (reads, explanations, typos, cosmetic
CSS, single config values) aligned with /hotfix scope.

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
bastien 2026-04-21 22:38:06 +02:00
parent 1da901ee8e
commit 31202e8621
1 changed files with 10 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
CLAUDE.md
View File

@ -28,7 +28,9 @@ Apply unless repo-specific instructions override.
3. If neither exists, create both before starting. 3. If neither exists, create both before starting.
## Workflow ## Workflow
- Non-trivial task (3+ steps): write plan in `tasks/TODO.md` first. Confirm before implementing. - Write/modify task touching logic (new behavior, control flow, state, API, dependencies): plan in `tasks/TODO.md` first, decomposed into subtasks. Task count doesn't matter — one complex task still requires a plan. When in doubt about complexity, skip the plan (be pragmatic).
- Confirm before implementing only when real trade-offs exist (multiple valid approaches, breaking change, destructive action) — otherwise proceed.
- Exempt from `TODO.md`: pure reads, explanations, questions, typos, cosmetic CSS tweaks, single config-value changes. Aligns with `/hotfix` scope (≤2 files, obvious fix).
- Minimal changes unless broader refactor requested. State trade-offs. - Minimal changes unless broader refactor requested. State trade-offs.
- Use sub-agents to keep main context clean — one task per sub-agent. Invest more compute on hard problems. - Use sub-agents to keep main context clean — one task per sub-agent. Invest more compute on hard problems.
- One question upfront if needed — never interrupt mid-task. *(Exception: orchestrators' mandatory validation gates — example, /init-project STEP 4/7, /ship-feature STEP 3 — are exempt.)* - One question upfront if needed — never interrupt mid-task. *(Exception: orchestrators' mandatory validation gates — example, /init-project STEP 4/7, /ship-feature STEP 3 — are exempt.)*
@ -45,10 +47,13 @@ Apply unless repo-specific instructions override.
5. After any correction: append to `tasks/LESSONS.md` — `[date] | what went wrong | rule to avoid it`. 5. After any correction: append to `tasks/LESSONS.md` — `[date] | what went wrong | rule to avoid it`.
## Task tracking (`tasks/TODO.md`) ## Task tracking (`tasks/TODO.md`)
1. Plan → write before implementing.
2. Confirm → explicit approval before starting. Applies to any write/modify task touching logic, regardless of task count. Skip for reads and trivial edits (see Workflow).
3. Track → mark done as you go.
4. Summarize → high-level summary at each major step. 1. Plan → write the task in `tasks/TODO.md` before touching code.
2. Decompose → split each complex task into subtasks (one subtask = one coherent change).
3. Track → check off subtasks as you go.
4. Summarize → high-level summary at each milestone.
## Context Navigation (graphify) ## Context Navigation (graphify)
- Use `/graphify query` ONLY for large-scope tasks: multi-file features, complex bug investigations, architectural changes, major refactors. - Use `/graphify query` ONLY for large-scope tasks: multi-file features, complex bug investigations, architectural changes, major refactors.