agent-environment-retr…
Use when a completed session needs an agent-environment retrospective. Not for an engineering retrospective from telemetry: use engineering-retrospective.
Use when asked to commit changes, create a typed branch, format history for a changelog, or rewrite messages of HEAD or an unpushed range. Not for pushing or a PR: use commit-push-pr.
$ npx -y skills add OutlineDriven/odin-claude-plugin --skill commit --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/commitContext preview
The summary Claude sees to decide when to auto-load this skill.
Use when asked to commit changes, create a typed branch, format history for a changelog, or rewrite messages of HEAD or an unpushed range. Not for pushing or a PR: use commit-push-pr.
name: commit description: 'Use when asked to commit changes, create a typed branch, format history for a changelog, or rewrite messages of HEAD or an unpushed range. Not for pushing or a PR: use commit-push-pr.'
Shape commit messages to repository conventions. Two modes: `create` (default) stages and commits working-tree changes as typed, atomic commits; `rewrite` rewords the messages of existing commits over `HEAD` or a short unpushed range while the tree stays byte-identical.
| Field | Bound contract | |---|---| | Trigger | Create: user asks to commit or save staged or unstaged changes, create a typed branch, or format history for changelog generation. Rewrite: user asks to clean up a finished commit message or bring messages of `HEAD` or a short unpushed range to handoff-ready form. | | Authority | Reversible local: create writes only the local index, a local feature branch, and local commits; rewrite writes only local commit messages, shows the exact commit set before mutation, and never creates or suggests a new commit; rollback is version control, and for rewrite the reflog and the original refs. No remote mutation. | | Side effect | Create: local working tree, local index, local branch refs, and local commit objects; rollback `git reset --hard <prior-HEAD>` discards new commits and `git branch -D <created-branch>` removes a branch this skill created. Rewrite: commit messages only; the tree stays byte-identical; each rewritten commit is re-signed with its author and committer dates preserved by position. Local history only. | | Done | Create: each logical concern is committed with a value-communicating message; `git status` confirms success; commit hashes and subjects are reported; history is parseable and auto-generates changelogs, SemVer bumps, and issue closes. Rewrite: `git diff <old>..<new>` is empty, each rewritten commit is signed with the intended dates by position, and `git log` alone carries the handoff with the diff hidden. |
1. Select the mode. `rewrite` when the user asks to clean up or reword existing commit messages; `create` otherwise. Done when: the mode is selected and its inputs are present or the missing one is reported. 2. Gather context: run `git status`, `git diff HEAD`, `git branch --show-current`, `git log --oneline -10`, and `git rev-parse --abbrev-ref origin/HEAD 2>/dev/null || echo DEFAULT_BRANCH_UNRESOLVED`. Strip the `origin/` prefix from the default branch. If it returned `DEFAULT_BRANCH_UNRESOLVED` or bare `HEAD`, try `gh repo view --json defaultBranchRef --jq '.defaultBranchRef.name'`; if that also fails, use `main`. Done when: working tree state, branch, recent history, and default branch are known. 3. Determine the commit message convention in priority order: (1) repo conventions in already-loaded project instructions, used directly; (2) a clear pattern in the recent commits from step 2 (conventional commits, ticket prefixes, emoji); (3) the seven-rule style: capitalized imperative subject, 50 chars target and 72 hard, no trailing period, blank line, body wrapped at 72 explaining what and why, never how. If source (1) or (2) uses conventional commits, format the subject as `<type>(<scope>): <subject>` (scope optional) and pick the most precise type from the allowed set with its SemVer effect: `feat` (MINOR), `fix` (PATCH), `docs`, `test`, `build`, `ci`, `chore`, `style` (formatting with no logic change), `ref` or `refactor` (refactoring with no behavior change), `perf`, `meta` (repository metadata), `license`, `revert` (none). Keep the subject ≤ 72 characters, imperative mood, no capital letter, no trailing period. Done when: the convention is determined and, if conventional, the type and SemVer effect are selected. 4. Mode rewrite (messages only; the tree stays byte-identical): 1. List the exact target set, `HEAD` or `HEAD~N..HEAD`, with `git log --format='%H %an <%ae> %ad %cn <%ce> %cd' <range>` so the set and per-position author and committer dates are visible. Done when: the exact commit set is listed with per-position dates. 2. Confirm none of the targets are pushed to a shared remote with `git branch -r --contains <sha>`. If any is pushed, stop and require explicit user confirmation before continuing; pushed history is otherwise untouched. Done when: all targets are confirmed unpushed or explicit confirmation is received. 3. Record each target's original tree SHA (`git rev-parse <sha>^{tree}`) and original author and committer dates, keyed by position in the range; re-signing must preserve these dates by position. Done when: every target's tree SHA and dates are recorded by position. 4. Rewrite each target's message to h
Formerly the ODIN Claude Plugin. The repository URL is unchanged. Outline-Driven Development, nicknamed ODIN, is a highly opinionated code-agent skill library: principles-first engineering, surgical editing, and workflow automation, published as installable
Repo: OutlineDriven/odin-claude-plugin
Use when a completed session needs an agent-environment retrospective. Not for an engineering retrospective from telemetry: use engineering-retrospective.
Use when a repo needs agent setup, AGENTS.md added or made lean, CLAUDE.md audited, or agent instructions scored or pruned. Not for remote, credential,…
Use when a human explicitly asks for a full repository agent-compatibility pass returning a scored report with prioritized fixes. Not for tasks that require…
Use when setting up a project, auditing agent command permissions, or asking which read-only bash commands and domains to allow. Not for remote, credential,…
Use when asked to build or review a CLI intended for coding agents and return flag-driven, pipeline-safe, idempotent design advice. Not for running or…
Use when the user asks to make the skills framework work in a new harness, IDE, or CLI. Not for remote, credential, publish, deploy, or irreversible changes.