filstawski/playbook

git-commit-messages

Conventional-commit-style message format — a type prefix (feat/fix/docs/ chore/refactor/...), an imperative present-tense summary under ~70 characters, an optional body explaining why rather than what, and no AI/assistant attribution or co-author watermark…

View source

How to use

  1. Copy the installation command and review the source package before installing it in your own agent environment.
  2. Check the source document for required tools, credentials and supported hosts.
  3. Describe your input and desired deliverable; review the output against the source material.
npx skills add https://github.com/filstawski/playbook --skill git-commit-messages

Before you start

  • Installation does not include a model subscription, API allowance or permission to access your files.
  • Repository stars and installation counts describe activity, not an independent quality or security test.
Original skill document

Rendered from the source repository. Headings, examples, code, tables, links, and referenced images are preserved.

Rule

  • First line: <type>(<optional scope>): <summary> — imperative

present tense ("add", not "added" or "adds"), under ~70 characters, no trailing period.

  • Type is whichever of these best matches the primary change:

feat, fix, docs, refactor, test, chore, build, ci, perf, style. If a change genuinely needs two types, that's two commits, not one commit picking the closer-sounding type.

  • Scope (the optional (scope) part) names the affected area only

when the repo is big enough that it actually disambiguates (feat(auth): ... in a large app). Skip it in a small repo where it adds no information over the summary alone.

  • Body (blank line, then prose) explains why the change was made —

the reasoning, the incident, the tradeoff behind it — not a restatement of the diff. The diff already says what changed; the body earns its place only by saying something the diff can't. Omit the body entirely for a change that's genuinely self-explanatory from the summary alone.

  • **No AI/assistant attribution lines, ever, unless explicitly asked for in

that instance.** No Co-Authored-By: Claude ..., no "Generated with ...", no session links, no watermark of any kind — regardless of what a tool's default commit template suggests appending. A commit's authorship metadata (the actual git author/committer fields) already records who ran the command; a text watermark in the message body is a separate, unwanted thing.

Examples:

feat: add per-route rate limit override

Some upload endpoints need a much higher ceiling than the global
default; per-route config keeps that from requiring a schema change.
fix: stop double-firing the delete audit entry
docs: correct the outdated symlink command in the README

Why

Conventional types make git log --oneline and any changelog tooling useful without extra setup, and a fixed format is one less thing to decide per commit. The body's job is specifically to carry information the diff itself cannot — motivation, context, a link to an incident — so a body that just narrates the diff in prose is wasted text nobody will read twice. Attribution/watermark lines are excluded on purpose: a commit is reviewed and authored under my direction as far as the repository's history is concerned, and I don't want that stamped as a permanent, unremovable annotation on every commit going forward just because a tool defaults to adding one.

Exceptions

  • A one-off WIP/checkpoint commit on a private branch that will be

squashed before merging doesn't need to follow this — conventional formatting is for commits that will actually live in shared history.

  • If a project's existing history already uses a different, established

convention (a team-wide scope list, a differently-shaped format), match that existing convention instead of imposing this one on a shared repo whose conventions aren't mine to unilaterally change.

See also

from this repository

More skills

All skills
filstawski
Community

api-manager-effects-and-populators

The two composition layers underneath the declarative manager pattern — deferred side-effects (realtime socket emit, audit log, CDN purge, all run independently via Promise.allSettled) and populators (batched, composable read-time enrichment, including the data-driven polymorphic feature table for share/like/comment/view-style cross-cutting features). Use when adding a new realtime event, wiring audit logging for a model, adding a derived/aggregate field to a manager's populated shape, or adding

1h installs
1
GitHub stars
0
Updated
Nov 3
filstawski
Community

code-polish

Pre-commit cleanup pass over the current change set: strip the debris left behind by trial-and-error iteration (debug logs, commented-out attempts, stacked workarounds, dead branches, defensive checks for impossible states, swallowed errors, any/ignore escapes), replace symptom patches with the one clean root-cause fix, then bring what remains up to professional standard — clear naming, doc comments on public API, idiomatic error handling, the project's own conventions, passing format/lint/typec

1h installs
1
GitHub stars
0
Updated
Nov 3
filstawski
Community

docs-migration-status-tracking

Track an incremental migration or a guideline-compliance backlog as an explicit checklist/status table in the docs, checked off as each piece actually completes, rather than trusting memory or a stale "done" claim. Use when migrating a system piece by piece (v1 to v2, old pattern to new pattern) where old and new coexist for a while, or when a written convention exists but the codebase hasn't fully caught up to it yet.

1h installs
1
GitHub stars
0
Updated
Nov 3