task-finish

task-finish is a skill for Claude Code, Codex from loncadev/baron. It costs 67 tokens per session (1,064 once invoked), scanned A, original, Apache-2.0.

A Baron workflow for finishing a work item by pushing its branch, opening or finding a draft pull request, and adding the pull-request link to the item.

In plain words
What is it for?
Prepare completed work for review from a clean work-item branch and connect the resulting pull request to its tracked item.
Why use it?
It prevents duplicate pull requests and ensures the work item points reviewers to the right code. The work item's role remains unchanged until the pull request is merged.

Skill for Claude CodeCodex

Part of the baron plugin — 10 skills shipped together

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/loncadev/baron/task-finish
Any agent
npx skills add loncadev/baron --skill task-finish
Clone the repo
git clone --depth 1 https://github.com/loncadev/baron

Made for: Claude Code, Codex.

Or install baron, the plugin that ships this one along with the rest of its 10 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for task-finish

README.md
[![agentmods](https://agentmods.dev/badge/skills/loncadev/baron/task-finish.svg)](https://agentmods.dev/skills/loncadev/baron/task-finish)
Your own site
<a href="https://agentmods.dev/skills/loncadev/baron/task-finish"><img src="https://agentmods.dev/badge/skills/loncadev/baron/task-finish.svg" alt="Measured on agentmods" height="20"></a>
Per session 67 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,064 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 $0.00067 $0.01064
Opus 5 $0.00034 $0.00532
Sonnet 5 $0.00013 $0.00213
Haiku 4.5 $0.00007 $0.00106

Measured 4d ago against content hash 0389168875d3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

task-finish 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 4d 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.

plugins/claude-code/skills/task-finish/SKILL.md · 69 lines

How it starts

The opening of the file, as written. The whole thing — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Finish a task (open the PR)

Idempotent finish: push the branch, check for an existing PR first, and only then run the task-finish recipe as ONE deterministic call. Local git is YOUR job; provider truth is Baron's.

Steps

  1. Identify the work item + branch: parse the current branch (git rev-parse --abbrev-ref HEAD) as <prefix>/<id>-<slug> and extract <id>. If the branch doesn't match, ask — never open a PR from a non-work-item branch without confirming.

  2. Preflight local git: git status --porcelain — dirty tree → stop (commit/stash first). Then push: git push -u origin HEAD (never --force; on a non-fast-forward reject, explain the rebase path and stop).

  3. Idempotency check — call baron_scm_read op=pr_for_branch with the branch:

    • PR exists (non-null): do NOT create another. Report its URL, and add a baron_scm_write op=pr_thread note only if there is genuinely new context (new commits since).
    • null: continue.
  4. Compose the PR description — never open an empty PR. A reviewer should not have to reconstruct the change from the diff. Write it from what you actually have: the work item (baron_issue_read op=get) plus the real commits/diff on the branch (git log <base>..HEAD --oneline, git diff --stat). Cover, briefly:

    • What changed — the substance, not a file list.
    • Why — the problem from the item; call out any decision you made along the way.
    • How it was verified — tests/build/manual check, and honestly what was NOT verified.

    Keep it short and specific; skip a section that has nothing real to say.

  5. Ask about auto-complete (AskUserQuestion, one question): "Merge automatically once checks pass?"Enable / No, I'll merge it myself. It is the user's call whether a PR lands unattended, so ask rather than assume — but ask ONCE and don't belabor it.

  6. Run the recipe — call baron_recipe_run exactly once:

    { "name": "task-finish", "inputs": { "issueId": "<id>", "branch": "<branch>", "title": "<PR title>", "body": "<the description you composed>", "autoComplete": true } }
    

Read the full file on GitHub · 69 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. 4d ago First seen · 69 lines · 67 tokens per session scan A 0389168875d3

Subscribe to this mod's changes

task-finish is a skill published in the GitHub repository loncadev/baron (1 stars, last pushed 13d ago), licensed Apache-2.0. It adds 67 tokens to every session and 1,064 once invoked, about $0.0003 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 2026-08-31.