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 nearform/lastlight --skill pr-reviewgit clone --depth 1 https://github.com/nearform/lastlightWrote 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/nearform/lastlight/pr-review)<a href="https://agentmods.dev/skills/nearform/lastlight/pr-review"><img src="https://agentmods.dev/badge/skills/nearform/lastlight/pr-review/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/nearform/lastlight/pr-review"><img src="https://agentmods.dev/badge/skills/nearform/lastlight/pr-review.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to high
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- high Tool Misuse · line 52 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
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 | $0.00069 | $0.04301 |
| Opus 5 | $0.00034 | $0.02150 |
| Sonnet 5 | $0.00014 | $0.00860 |
| Haiku 4.5 | $0.00007 | $0.00430 |
Grade A, and why
pr-review 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 9d 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 — 318 lines — stays where its author put it; the contents beside it link to each section on GitHub.
PR Review
Review an open PR — high-signal findings only. Read the change and reason about it; where reasoning cannot settle a question, run something. Installing the repo's dependencies, opening the installed library source, and writing the smallest file that exercises the behaviour and executing it are all allowed and expected — that is a probe, and it is how a question about how code actually behaves gets settled instead of guessed.
Two limits on a probe, and they are what keep it from becoming a second CI:
- It must produce evidence you can quote. Keep the command and its output, and cite them in the finding that rests on them. "I ran it and it fails" with nothing to quote is worth exactly what a guess is worth.
- It is targeted at one question, never a re-derivation of CI. Whether the change builds and whether the suite is green are already answered — see §4. Never spend a probe on those.
Why this is spelled out rather than left implicit. An earlier version of this skill forbade installing dependencies, and the measured failure was not disobedience — it was the opposite. The reviewer referred to
WebClient32 times and never once openednode_modules/@slack, because the workspace it was given had nonode_modulesat all. "Open the library source" was not ignored; it was structurally impossible. An affordance you do not have reads to you as an instruction you cannot follow, so what you can run is part of the contract, not an implementation detail.
Your job is judgement on the diff, not a build gate. A noisy review gets muted, so precision matters more than volume.
You do not post the review yourself. You write your findings to a JSON file
(.lastlight/pr-review/findings.json) and a deterministic follow-up step posts
one formal review, anchoring each finding to its diff line as an inline comment
(§5).
This skill is the PR-specific procedure. It uses the code-review skill for the precision bar and what-to-check rubric.
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.
- 9d ago First seen · 318 lines · 69 tokens per session scan A 3cc006f7ba15
pr-review is a skill published in the GitHub repository nearform/lastlight (22 stars, last pushed yesterday), licensed MIT. It adds 69 tokens to every session and 4,301 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 2026-08-30.
Other skills, from other repositories
代码审查员
A code-review agent that examines changes for correctness, security, maintainability, performance, and test coverage.
Codex
Guides operator-owned delegation to Codex when the task fits a bounded coding lane, review pass, or worktree-isolated implementation track.
open-code-review-delegate
Delegation mode for open-code-review (OCR). Instead of OCR calling an LLM endpoint, this skill instructs the host agent to perform the code review itself, using OCR only for deterministic engineering: file selection and rule resolution. Use when the host agent should drive the review with its own LLM capabilities.
open-code-review
Performs AI-powered code review on Git changes using the ocr CLI from alibaba/open-code-review. Use when the user asks to review code, review a pull request, review staged/unstaged changes, review a commit, or compare branches for code quality issues. Produces line-level review comments and can automatically apply…
chrome-cdp
Drive a headless Chrome over the Chrome DevTools Protocol (CDP) for browser QA — navigate, click, fill forms, read the DOM/accessibility tree, screenshot, and assert. Use whenever a task requires loading a web page and interacting with it like a user. Chrome is launched by a bash step (recipe below); this skill…
claude_code_delegation
Delegate coding, review, and refactor tasks to Claude Code CLI through Row-Bot's approval-gated shell workflow.