brainstorming
Use before coding when requirements have multiple reasonable interpretations, approaches require tradeoffs, or mistakes would be costly…
Use in cross-session work when user-confirmed outcomes, non-goals, hard constraints, or acceptance criteria have not been written back to the current source of truth (spec), or reality conflicts with it enough that following it would be wrong; also use during wrap-up or handoff
$ npx -y skills add chipfighter/coding-discipline --skill spec-sync --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/spec-syncContext preview
The summary Claude sees to decide when to auto-load this skill.
Use in cross-session work when user-confirmed outcomes, non-goals, hard constraints, or acceptance criteria have not been written back to the current source of truth (spec), or reality conflicts with it enough that following it would be wrong; also use during wrap-up or handoff
name: spec-sync description: Use in cross-session work when user-confirmed outcomes, non-goals, hard constraints, or acceptance criteria have not been written back to the current source of truth (spec), or reality conflicts with it enough that following it would be wrong; also use during wrap-up or handoff when a source of truth should exist but does not. Edit only the affected original wording; do not keep separate old and new versions. Do not trigger for work that can be completed within one session, local implementation details, verification against existing acceptance criteria, locating the source of truth, or ordinary documentation edits. license: MIT
Hard rule: **A change confirmed in conversation must not remain only in the conversation.** If user-confirmed intent that affects future collaboration or acceptance is not written back to the source of truth in this session, the next session cannot even see that the user changed their mind—intent that was never persisted cannot be recovered.
> If the next session or another agent read only the current source of truth, would it continue toward an outcome the user has rejected, violate a new constraint, or judge completion against stale criteria? If so, the change must be synced.
1. Use the current source of truth already designated by the project—the document named by the project guide. 2. If none is designated, designate an existing Issue / PRD / ADR / status document as the current spec; **do not copy it into a second document**. If there are multiple candidates or none is obvious, ask the user to choose. 3. If nothing exists and the work genuinely spans sessions, is shared by multiple people, or would be costly to get wrong, ask whether the user wants a minimal spec (below). The user decides whether to create one; do not create one for work that can be completed within a single session. 4. After designating or creating one, put a pointer to the source of truth in the project guide document (the “where to find the current state” entry in `CLAUDE.md` / `AGENTS.md`)—**write only the pointer, not a copy of the content**—so the next session can find it. 5. **Edit only the affected original wording**: do not append separate “old decision / new decision” versions, reorder the document, or polish the whole document along the way. Replace the affected text as a whole only when the direction changes substantially; leave history to Git. 6. Do not record execution progress, next steps, or verification evidence. Those belong to the harness / Git / CI, not the spec. 7. In multi-person / multi-agent work, workers are read-only by default; the primary agent or owner maintains the spec.
The current desired outcome; explicit non-goals; hard constraints; acceptance criteria for this work; confirmed unresolved questions that would block safe execution (include only questions the user has confirmed—do not promote your guesses into settled facts).
Which document is the current source of truth and how to read it belong to context-hygiene. Verifying completion against existing acceptance criteria belongs to verify-before-done. This skill governs one thing only: **whether the source of truth reflects the user's latest confirmed decisions.**
Chinese → README.zh-CN.md Guardrails, not a workflow. Quiet on routine work; firm when a named failure mode is at risk. AI coding agents tend to fail in two expensive ways: They drift off-goal.
Repo: chipfighter/coding-discipline
Use before coding when requirements have multiple reasonable interpretations, approaches require tradeoffs, or mistakes would be costly…
Use when changes affect interactions across multiple modules, involve high-risk areas (authorization/authentication, payments/funds, data deletion/migration,…
Use when starting a project, reading project documentation or history, or resolving conflicts between documents. For current state, read only the latest…
Use when creating a branch, using worktrees for parallel work, committing, or wrapping up. Covers branch naming, conventional commits, short-lived branches,…
Use before proposing a permanent fix for bugs, test failures, or unexpected behavior with an unknown root cause—mandatory when symptoms are far from the cause,…
Use before implementation when behavior can be verified by automated tests and regression coverage has clear value. For bugs with an unknown root cause, use…