agentleFS
lutzleonhardtlutzleonhardt/casefile/skills/commit/SKILL.md

commit

Skill4 starsChanged 59 days agoMITView on GitHub

Commit a task's code and its wrap-up summary

188 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use validate-plan, act-on-feedback and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • shawn-dumasSkill

    validate-plan

    Validate an orchestration plan after pre-flight certification. Runs adversarial review, deep review (verify prompt data against codebase), and PoC gate for risky approaches. Auto-invoked by orchestrate-* skills at Step 8.

    Also in 5 projects that use this

  • MicrosoftSkill

    act-on-feedback

    Act on user feedback attached to the current session. Use when the user submits feedback on the session's changes via the Submit Feedback button.

    Also in 3 projects that use this

  • decocmsSkill

    add-mcp-tools

    Guide for adding new MCP tools with consistent patterns for schemas, tool definitions, registry updates, and Better Auth integration

    Also in 3 projects that use this

  • MicrosoftSkill

    generate-run-commands

    Generate or modify run commands for the current session. Use when the user wants to set up or update run commands that appear in the session's Run button.

    Also in 3 projects that use this

  • mitsuhikoSkill

    oebb-scotty

    Austrian rail travel planner (ÖBB Scotty). Use when planning train journeys in Austria, checking departures/arrivals at stations, or looking for service disruptions. Covers ÖBB trains, S-Bahn, regional trains, and connections to neighboring countries.

    Also in 3 projects that use this

  • mitsuhikoSkill

    pi-share

    Load and parse session transcripts from shittycodingagent.ai/buildwithpi.ai/buildwithpi.com/pi.dev (pi-share) URLs. Fetches gists, decodes embedded session data, and extracts conversation history.

    Also in 3 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. 0 are common, 6 are distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 168 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If <work-root>/plan.md is missing, stop and tell the user to run /plan on this branch first.

    line 45
  2. Very specific:

    Link: after the code commit, one command attaches the git note (tasklog: pointer), commits the casefile repo, and pushes the notes backup …

    line 58
  3. Very specific:

    Use $ARGUMENTS to identify the task being committed (e.g.

    line 106
  4. Very specific:

    If $ARGUMENTS contains a number, use that.

    line 111
  5. Distinctive:

    Do not guess. Do not run git commands.

    line 116
  6. Distinctive:

    Do not auto-resolve any of these — surface them and wait.

    line 178
  7. Very specific:

    Every file in the log's Files Modified section that exists in the working tree or index.

    line 185
  8. Very specific:

    Keep messages short (≤ 72 chars for the subject line).

    line 225
  9. Distinctive:

    When a body is warranted, keep it short and traceable:

    line 229
  10. Very specific:

    Prefer Covers over Closes.

    line 237
  11. Distinctive:

    Commit message (subject line).

    line 245
  12. Distinctive:

    Do not push. Pushing is an explicit human action.

    line 292
  13. Distinctive:

    Every path in the staging list is explicit and traceable to the log.

    line 303

“Common” is measured: kept by at least 6 in 10 of the 168 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Task Commit

The task's wrap-up summary already exists (written by /wrap-up N). Stage and commit the task's code changes together with the summary file in a single commit.

For a task whose change set reads better as several small commits, /commit-split N is the optional alternative: same log, same confirmation, a short series of slices cut by intent, each verified green on its own.

This skill executes git add and git commit after showing the user the full plan and waiting for explicit confirmation. If the user says anything other than a clear "yes" / "ok" / "commit", abort without running any git command.

Work scope

Resolve the active roots before touching any plan, log, or doc path:

casefile root -v

Trust its output verbatim: work root is where plan.md and task-log/ live — every docs/work/<scope>/ path in this skill refers to it. doc root is where project-global docs (specs, architecture, improvements) live — every other docs/ path in this skill resolves against it. scope and mode come from the same output; do not re-derive any of them, and let the CLI's own errors (detached HEAD, missing casefile repo) stop the run. If casefile is not on PATH, stop: the kit ships with its CLI — reinstall it instead of deriving paths by hand. In casefile mode, work artifacts never enter the current repository. archive is where casefile archive copies session transcripts; it is never a target for plans, logs, or docs.

If <work-root>/plan.md is missing, stop and tell the user to run /plan on this branch first. Legacy layouts (docs/plans/, a root plan.md, docs/task-log/) are never automatic fallbacks — migrate them into the scoped work root first.

Casefile mode differences

In casefile mode the work artifacts never enter the current repository. This changes the commit mechanics:

  • Staging list: [code] files only. The [log] file (and [plan] for BLOCKED+replan) live in the casefile repo instead, not staged here.
  • Link: after the code commit, one command attaches the git note (tasklog: pointer), commits the casefile repo, and pushes the notes backup into the casefile:
    casefile link {N} {slug} [{sha}] -m "task-N: <title> ({project})"
    By default this also copies the transcripts listed in the log's ### Sessions section into the casefile (safety net against transcript expiry and machine loss). Pass --no-session only when the user explicitly wants the raw transcripts excluded.
  • Code already committed (e.g. wrap-up ran after the commit): skip git add/git commit, ask the user to confirm the target commit hash, then run casefile link {N} {slug} <hash> against it.
  • Never push notes refs or anything else to the repository's origin. casefile link's backup push targets the private casefile only.

The casefile link command appears in the commit plan (step 5) and runs only after the user confirms.

Home mode: session archive

In home mode the log and code are committed to the repository itself — but the session transcripts under ~/.claude/projects/ (and ~/.codex/sessions/) are ephemeral: nothing preserves them. /commit therefore archives them into the casefile repo, which acts as the personal session archive even for home-mode projects. After the code commit, run:

casefile archive {N}

It copies the transcripts listed in the log's ### Sessions section plus the current session (a resume's ID may not be in the log yet — the commit phase always runs after the last wrap-up) into <casefile>/<project>/work/<scope>/sessions/task-{N}/ and commits the casefile repo. If it fails with "no casefile repo", report that one line and continue — the archive is a safety net, never a commit blocker. Missing source files (expired transcripts) are reported per line, not fatal. Skip the step only when the user explicitly asks for the transcripts to be excluded.

The command appears in the commit plan (step 5, [sessions] group) and runs only after the user confirms, once the code commit has succeeded.

Task identity — $ARGUMENTS

Use $ARGUMENTS to identify the task being committed (e.g. /commit 3 means Task 3). Required.

Resolution rules, in order:

  1. If $ARGUMENTS contains a number, use that.
  2. Else, if this session ran /wrap-up N or /start-task N and N is unambiguous in context, use N.
  3. Else, stop and tell the user:

    Task number required. Run as /commit N. Do not guess. Do not run git commands.

Fix lane: $ARGUMENTS may also name an unplanned work item (see /wrap-up's fix lane). A full stem (fix-<slug>, chore-<slug>, …) is used as-is. A bare type (fix) resolves via (a) this session's /wrap-up fix identity, else (b) exactly one uncommitted <type>-*.md log in the scope's task-log/ — if several match, list them and stop. Everywhere this skill says task-{N}-{slug}.md, a fix item reads {stem}.md; everywhere it says {N} (including casefile link and casefile archive), pass the stem — casefile link fix-<slug> [sha] takes no separate slug argument.

Workflow

1. Locate the log file

ls docs/work/<scope>/task-log/task-{N}-*.md
  • Exactly one file — continue.
  • No file — stop. Tell the user:

    No wrap-up summary found for Task N. Run /wrap-up N first.

  • Multiple files — stop. Filename convention violated inside this work scope; ask the user to consolidate before committing.

2. Read the log file

Extract:

  • Title (from ### Task — the one-sentence summary).
  • Status (DONE or BLOCKED).
  • Files Modified list — the canonical list of source files that should be staged.
  • Acceptance Coverage — note covered, partial, skipped, or deferred AC IDs for the optional commit body.
  • For BLOCKED: note the Re-Plan Proposal if present and identify the plan file path referenced.

3. Check current git state

git status --short
git diff --stat

Reconcile against the Files Modified list:

  • Files listed in the log that are missing from the working tree / index → flag to user. Possible causes: file was reverted, already committed separately, or log is stale.
  • Files present in the working tree that are not in the log → flag to user. Either the log is incomplete (re-run /wrap-up N to refresh) or these files belong to a different task.
  • Files already staged that are not in the log → flag. The user might have staged something by hand; ask before sweeping it into the task commit.

Do not auto-resolve any of these — surface them and wait.

4. Build the staging list and commit message

Staging list:

  • The log file: docs/work/<scope>/task-log/task-{N}-{slug}.md
  • Every file in the log's Files Modified section that exists in the working tree or index.
  • For BLOCKED with Re-Plan: also include docs/work/<scope>/plan.md (the updated plan).

Casefile mode: stage [code] files only — the log and plan are committed to the casefile repo instead (see Casefile mode differences).

Commit message template (home mode):

  • DONE: task-N: <title from log>
  • BLOCKED with re-plan: replan: <reason> (task-N blocked) The <reason> is a short human-readable phrase derived from the Re-Plan Proposal — ask the user if ambiguous.
  • BLOCKED without re-plan: task-N: blocked — <reason>
  • Fix lane: <type>: <title from log> (e.g. fix: …, chore: …) — the stem's type is the conventional-commit type.

Commit message template (casefile mode): the code repo does not carry the plan or task logs, so workflow vocabulary (task-N, replan, blocked) must not leak into its history — to colleagues it references a numbering that does not exist. Use a plain, conventional message instead:

  • DONE: <title from log> (no task-N prefix)
  • BLOCKED: a neutral WIP-style message derived from the title — ask the user if unsure what is appropriate in this repo.

The task number stays fully traceable: it lives in the git note (tasklog: .../task-{N}-{slug}.md) and in the casefile repo's own commit message, which keeps the task-N: prefix.

The same leak rule covers the diff itself: in casefile mode, AC IDs (T{N}-AC-{NN}, XC-NN) and task numbers must not appear in source comments, test names, or fixtures of the code repo. Before building the commit plan, git grep -nE 'T[0-9.]+-AC-[0-9]+|XC-[0-9]+' over the [code] files; any hit is a discrepancy in step 5 for the user to decide on. Home mode is exempt — there the plan sits next to the code.

Keep messages short (≤ 72 chars for the subject line). No automatic body unless the user asks for one or the Acceptance Coverage section contains anything other than passed.

When a body is warranted, keep it short and traceable:

Covers T{N}-AC-01..05
Partial T{N}-AC-06: <reason>
Defers T{N}-AC-07 -> Task M

Prefer Covers over Closes; AC IDs are not issue IDs. In casefile mode the code-repo commit gets no such body at all — the Acceptance Coverage is readable in the linked task log.

5. Show the commit plan

Present a single block to the user containing:

  • Commit message (subject line).
  • Files to be staged, one per line, grouped as:
    • [log] — the task summary file (home mode only)
    • [code] — source files from Files Modified
    • [plan] — plan file (BLOCKED+replan only, home mode only)
    • [casefile] — casefile mode: the exact casefile link … command
    • [sessions] — home mode: the exact casefile archive … command (see Home mode: session archive)
  • Any discrepancies from step 3 (missing / extra / already-staged files) with a short note each.
  • The exact commands that will run:
    git add <file1> <file2> ...
    git commit -m "<message>"
    # or: git commit -m "<message>" -m "<body>" when a body is used

End with a confirmation prompt, e.g.:

Commit Task N as shown above? (yes / no / edit message)

6. Act on the response

  • yes / ok / commit — run git add <files> then git commit -m "<message>" (plus -m "<body>" if the shown plan included a body). Show the resulting commit hash and a one-line confirmation. In casefile mode, then run the [casefile] command from the plan against the new hash and confirm with one line. In home mode, then run the [sessions] command from the plan and confirm the copy count and casefile commit with one line.
  • edit message — accept a revised message from the user and re-show the plan for confirmation. Do not commit until re-confirmed.
  • no or anything ambiguous — abort. Do not stage, do not commit. Tell the user nothing was changed.

7. After a successful commit

Tell the user:

Committed as <hash>. Task N is closed.

Next: /start-task N+1 to begin the next task, or /review for a quick per-task review, or /review full before opening a PR.

Do not push. Pushing is an explicit human action.

What this skill does not do

  • It does not run tests. Test evidence lives in the wrap-up.
  • It does not amend or rewrite history. If the user needs to amend a prior commit, they do that by hand.
  • It does not push to a remote. (Sole exception: casefile link's notes backup push, which targets the local casefile path and never the repository's origin.)
  • It does not sweep files with git add -A or git add .. Every path in the staging list is explicit and traceable to the log.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.