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
- daledesilva/obsidian_inktheir own version, first seen Nov 9, 2026
- thoughtbot/rails-consultanttheir own version, first seen Nov 8, 2026
- pedrohcgs/claude-code-my-workflowtheir own version, first seen Nov 8, 2026
- jisuzzzz/haechitheir own version, first seen Nov 8, 2026
- dirien/yet-another-agent-harnesstheir own version, first seen Nov 8, 2026
- denysdovhan/skillstheir own version, first seen Nov 8, 2026
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-dumasSkillvalidate-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
MicrosoftSkillact-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
decocmsSkilladd-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
MicrosoftSkillgenerate-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
mitsuhikoSkilloebb-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
mitsuhikoSkillpi-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
Other versions188 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 188
- microsoft/vscode★ 195ksame name
- TryGhost/Ghost★ 56ksame name
- ToolJet/ToolJet★ 41ksame name
- ShareX/ShareX★ 41ksame name
- openai/symphony★ 27ksame name
- ccusage/ccusage★ 19ksame name
- ryoppippi/ccusage★ 19ksame name
- windmill-labs/windmill★ 18ksame name
- EverMind-AI/EverMemOS★ 14ksame name
- EverMind-AI/EverOS★ 14ksame name
178 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository9 other files it gives its agents.
- commit-splitSkill
- csSkill
- finalize-squashSkill
- handoffSkill
- planSkill
- reviewSkill
- start-taskSkill
- whySkill
- wrap-upSkill
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
- Very specific: line 45
If <work-root>/plan.md is missing, stop and tell the user to run /plan on this branch first.
- Very specific: line 58
Link: after the code commit, one command attaches the git note (tasklog: pointer), commits the casefile repo, and pushes the notes backup …
- Very specific: line 106
Use $ARGUMENTS to identify the task being committed (e.g.
- Very specific: line 111
If $ARGUMENTS contains a number, use that.
- Distinctive: line 116
Do not guess. Do not run git commands.
- Distinctive: line 178
Do not auto-resolve any of these — surface them and wait.
- Very specific: line 185
Every file in the log's Files Modified section that exists in the working tree or index.
- Very specific: line 225
Keep messages short (≤ 72 chars for the subject line).
- Distinctive: line 229
When a body is warranted, keep it short and traceable:
- Very specific: line 237
Prefer Covers over Closes.
- Distinctive: line 245
Commit message (subject line).
- Distinctive: line 292
Do not push. Pushing is an explicit human action.
- Distinctive: line 303
Every path in the staging list is explicit and traceable to the log.
“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.
Files
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 -vTrust 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:
By default this also copies the transcripts listed in the log'scasefile link {N} {slug} [{sha}] -m "task-N: <title> ({project})"### Sessionssection into the casefile (safety net against transcript expiry and machine loss). Pass--no-sessiononly 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 runcasefile 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:
- If
$ARGUMENTScontains a number, use that. - Else, if this session ran
/wrap-up Nor/start-task Nand N is unambiguous in context, use N. - 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 Nfirst. - 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 (
DONEorBLOCKED). - 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 --statReconcile 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 Nto 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 Modifiedsection 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 MPrefer 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 exactcasefile link …command[sessions]— home mode: the exactcasefile 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— rungit add <files>thengit 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.noor 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+1to begin the next task, or/reviewfor a quick per-task review, or/review fullbefore 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 -Aorgit add .. Every path in the staging list is explicit and traceable to the log.
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
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.

