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 nikeyes/stepwise-dev --skill review-pr-commentsgit clone --depth 1 https://github.com/nikeyes/stepwise-devWrote 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/nikeyes/stepwise-dev/review-pr-comments)<a href="https://agentmods.dev/skills/nikeyes/stepwise-dev/review-pr-comments"><img src="https://agentmods.dev/badge/skills/nikeyes/stepwise-dev/review-pr-comments/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/nikeyes/stepwise-dev/review-pr-comments"><img src="https://agentmods.dev/badge/skills/nikeyes/stepwise-dev/review-pr-comments.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00026 | $0.01011 |
| Opus 5.5 | $0.00010 | $0.00404 |
| Sonnet 5.5 | $0.00005 | $0.00202 |
| Haiku 4.5 | $0.00003 | $0.00101 |
Grade A, and why
review-pr-comments 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 8d 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 — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Review PR Comments
Review the pending comments on a PR, decide on each one, implement the accepted changes, push them and reply in every thread. Do not ask for confirmation: the report at the end is how the user learns what you decided.
Rules
- PR target: the PR number if given, otherwise the PR of the current branch. If there is none, say so and stop.
- Plan: if a plan file is given, read it. It is the agreed design: a comment that asks to go against it needs a stronger reason to be accepted.
- Scope: comments from anyone, bots and people alike. "Yours" below means written by the authenticated GitHub user.
- Pending:
- An inline thread is pending when it is not resolved, not outdated, and its last comment is not yours.
- A general comment or a review body is pending when it is not yours and was created or edited after your last comment on the PR. If you never commented, all of them are.
- If nothing is pending, say so and stop.
- Bias: defend the existing code. Only accept a change with a concrete technical reason (bug, violated principle, demonstrable readability gain). Reject style preferences without objective justification.
- Context: before evaluating, read the full files affected plus related files (tests, types, direct dependencies).
- Severity: take it from the comment's prefix when present (
[SECURITY — CRITICAL],[CODE QUALITY — HIGH],[TESTS — LOW],[SUGGESTION], etc.). Otherwise infer it: correctness or security bugs are HIGH; performance, tests or missing error paths are MEDIUM; style, docstrings, naming and "for consistency" nits are LOW. - Decision: one of these for every pending comment.
ACCEPT: you will make the change.REJECT: you will not, and you can say why.REPEATED: it raises a point you already answered on this PR and adds no new argument. Do not reply again.
- Implementing: change only what the accepted comments ask for.
- A change in behavior goes test-first. If
/stepwise-core:tddis available, invoke it with theSkilltool: describe the accepted comment as the behavior to implement and pass the plan file if you have one. If it is not available, write the failing test yourself before the fix. - A change that does not alter behavior (naming, comments, docs, configuration) is a direct edit.
- If you cannot make an accepted change, do not pretend: reply saying what blocked it and list it in the report.
- A change in behavior goes test-first. If
- Publishing the code: if any file changed, invoke
/stepwise-git:commitwith theSkilltool, thengit push. Never force-push. Do this before replying. If the commit or the push fails, do not post any reply that says a change is done: stop and report. - Replying: one reply per
ACCEPTand perREJECT, posted after the code is pushed.- Inline threads get a reply in the thread. General comments and review bodies get a general PR comment that links to the comment it answers.
ACCEPT: what changed and the commit that changed it.REJECT: the technical reason, and what you propose instead if anything. One line is enough for LOW.- Always in English, whatever the comment's language. No Claude attribution.
- Never resolve a thread: that belongs to whoever opened it.
- Report: every comment grouped by severity, one block each, and then whether you pushed code (with the commits), so a caller knows a new round of checks has started.
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.
- 8d ago Changed · -6 lines · +6 tokens per session d6c9c5e3fec3
- 1mo ago Changed 67914dd4dfb8
- 1mo ago Changed 0d0298ac1046
- 1mo ago First seen · 56 lines · 20 tokens per session scan A f985e6144990
review-pr-comments is a skill published in the GitHub repository nikeyes/stepwise-dev (24 stars, last pushed 6d ago), licensed Apache-2.0. It adds 26 tokens to every session and 1,011 once invoked, about $0.0001 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 2026-08-30.
Other skills, from other repositories
autoreview
Structured code review through the shared OpenClaw agent-skills installation.
omh-code-review
This is an OMH code-review workflow skill, projected for Agent Skills hosts (Claude Code, Codex, Cursor, opencode, OpenClaw, pi).
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
asplos-writing-style
Use when drafting or revising ASPLOS prose — engineering the first two pages that rapid review actually reads, stating the cross-layer insight as one quotable sentence, structuring per-layer mechanism sections, writing bounded claims, and fitting everything into 11 self-contained pages that include figures and…
explore-unknowns
Guide the user through a quadrant walk that maps the unknowns of a task — open by listing the known knowns, then work through known unknowns, unknown knowns, and unknown unknowns one stage at a time, ending with a complete four-quadrant map in the user's hands. Use when a request is ambiguous or underspecified, the…
re0-merge
Review and land an external contribution the way this suite does: gate it against the thesis, land it with the author's credit intact, complete a new skill rather than merging it raw, then approve, credit, and explain before closing. Use when reviewing a pull request, as any collaborator or maintainer, not only the…