commit-work

A skill for checking, staging, and committing a reviewed coding work step with a focused conventional commit.

In plain words
What is it for?
It helps finalize approved changes, run required checks, stage only relevant files, and create a properly labelled Git commit.
Why use it?
It prevents unfinished plans, unresolved review findings, stale documentation, or unrelated files from being committed.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/jakubsuplicki/codument/commit-work
Any agent
npx skills add jakubsuplicki/codument --skill commit-work
Clone the repo
git clone --depth 1 https://github.com/jakubsuplicki/codument

Made for: Claude Code, Codex.

Per session 21 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 814 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00021 $0.00814
Opus 5 $0.00010 $0.00407
Sonnet 5 $0.00004 $0.00163
Haiku 4.5 $0.00002 $0.00081

Measured 2d ago against content hash cca1c41bebc9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

commit-work scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.agents/skills/commit-work/SKILL.md · 49 lines

How it starts

The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Commit Work

Use this after review-work is clean, or after the user has approved/deferred every review finding and all approved fixes are resolved.

Workflow

  1. Check the active plan step is complete. If it was the final step, confirm the ## Delivery Plan block has been compacted out (plan-with-docs → Compaction on ship) — a shipped feature doc must not commit with a stale delivery checklist still in it.
  2. Check that review-work is clean, or that every finding was fixed or explicitly deferred by the user.
  3. Check git status --short.
  4. Review the diff and avoid staging unrelated user changes.
  5. Run the relevant verification commands.
  6. Run codument review --strict — it must pass. It exits nonzero while the step left a new source unmapped or a mapped doc stale; materialize the file(s) (codument map materialize <file>) and update the stale doc(s), then re-run until clean. Do not commit while it is red.
  7. Stage only files belonging to the completed step.
  8. Commit with a conventional commit prefix:
    • feat:
    • fix:
    • docs:
    • test:
    • refactor:
    • chore:
  9. If that was the plan's last step, run codument doctor --strict once and report what it says. It reports; it does not gate — a plan must not be blocked by an adopting repo's pre-existing debt. This is the only moment the loop looks at repo-wide health: review answers "is this change in sync", doctor answers "is the knowledge base still worth reading", and a field run made 37 review-family calls across five commits without ever learning its own health surface stood at 140 findings, including one on a doc that session had just edited. Once per plan, not once per step.
  10. Continue directly to work-step for the next unchecked step, or report completion if none remain. In gated mode, stop after the commit and offer the next-step gate instead:
Step N is reviewed and committed. Next options:
1. Start the next unchecked plan step with /work-step
2. Review the plan before continuing
3. Compact context before continuing
4. Pause here
  1. If the user chooses compact context, use the active agent's native context-compaction command when one is available. If no native command is available, provide a concise restart note grounded in AGENTS.md, the active plan doc, docs/.registry.json, and git status, then pause.

Read the full file on GitHub · 49 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 49 lines · 21 tokens per session scan A cca1c41bebc9

Subscribe to this mod's changes

commit-work is a skill published in the GitHub repository jakubsuplicki/codument (47 stars, last pushed 12d ago), licensed Apache-2.0. It adds 21 tokens to every session and 814 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.