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.
npx skills add mthines/agent-skills --skill review-branchgit clone --depth 1 https://github.com/mthines/agent-skillsWrote 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.
[](https://agentmods.dev/skills/mthines/agent-skills/review-branch)<a href="https://agentmods.dev/skills/mthines/agent-skills/review-branch"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/review-branch/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/mthines/agent-skills/review-branch"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/review-branch.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00222 | $0.03645 |
| Opus 5.5 | $0.00089 | $0.01458 |
| Sonnet 5.5 | $0.00044 | $0.00729 |
| Haiku 4.5 | $0.00022 | $0.00364 |
Grade A, and why
review-branch 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 16d 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.
How it starts
The opening of the file, as written. The whole thing — 275 lines — stays where its author put it; the contents beside it link to each section on GitHub.
review-branch — local convergence, no PR required
Drive a branch to a clean state by iterating review → apply → simplify → re-review, with every finding carried in a local file instead of a GitHub review thread.
This is review-loop with the GitHub round trip removed. The findings
are the same findings — the branch-reviewer agent runs the
same detection core pr-reviewer does, by reference rather than by copy — but nothing is posted,
nothing is fetched, and the loop's state lives in
.agent/{branch}/findings.jsonl.
When to use which:
review-branch |
review-loop |
|
|---|---|---|
| Needs a PR | no | yes |
| Bus | local findings.jsonl |
GitHub review threads |
| GitHub calls | zero | many |
| Green gate | the repo's own fast checks | CI |
| Right when | iterating on your own work, pre-PR, offline | a PR exists and other people read the threads |
The thread is not overhead once someone else is reading it — it is the record, and it is how a human
reviewer sees what the agent already handled. Reach for review-loop the moment a PR exists.
Modes
| Flag | Effect |
|---|---|
--base <ref> |
Base of the comparison. Default: upstream → origin/HEAD → main, and the loop announces which rung answered. |
--head <ref> |
Head of the comparison. Default: the working tree, so uncommitted work is reviewed. |
--cap N |
Iteration cap. Default 5. |
--effort high |
Forces the deep tier and raises diversify-then-vote from 3 finders to 5. |
--no-simplify |
Skip sub-step C (polish simplify). |
--no-checks |
Skip sub-step D. Convergence then means findings-clean only, and the report says so. |
--report |
Report-only. Forces CAP=1, skips B/C/D, applies nothing. The local counterpart of review-loop's --no-feedback. |
--include-untracked |
Treat untracked, non-ignored files as added. Off by default. |
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 16d ago Changed · -1 lines f37de693f950
- 19d ago Changed 1a2c0ce80891
- 1mo ago First seen · 276 lines · 222 tokens per session scan A e064cb0ba4fb
review-branch is a skill published in the GitHub repository mthines/agent-skills (19 stars, last pushed today), licensed MIT. It adds 222 tokens to every session and 3,645 once invoked, about $0.0009 per session on Opus 5.5. 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 2026-09-13.
Other skills, from other repositories
local-pr-review
Act as a senior engineer and review the committed changes on the current git branch against the base branch (develop or main) — a local, pre-push PR review. Use this skill whenever the user wants feedback on a branch as a whole before opening or merging a pull request — phrases like "review my branch", "PR review…
local-review
Act as a senior engineer and review all uncommitted changes in the local git branch — both staged and unstaged, including new untracked files — before they get committed. Use this skill whenever the user wants pre-commit feedback on work in progress — phrases like "review my changes", "review my diff", "code review…
raise-pr
Push the current feature branch and raise a GitHub pull request with a conventional-commit style title and a structured body, after verifying the branch actually has committed work to raise. Use this skill whenever the user asks to "raise a PR", "open a pull request", "create a PR", "push and PR", "get this reviewed"…
check
Reviews diffs, PRs, release readiness, and publishing follow-through. Use when asked to review, triage issues or PRs, or ship. Not for debugging root causes or prose.
ego-lite-simplify
Find and implement evidence-backed simplifications in the ego-lite repository. Use when reviewing the codebase for dead code, duplicated state or APIs, speculative abstractions, unnecessary compatibility layers, hand-written infrastructure, excessive tests or documentation, or when the user asks to reduce code size or…
antfu-create-pr
Create a reviewable GitHub pull request from the current branch with a Conventional Commits title, a concise evidence-based body, and before/after screenshots for UI changes. Use when asked to open, create, publish, or prepare a PR.