For open source maintainers

Review like a
maintainer

Eight agent skills that do the slow part of reviewing a pull request — the careful reading — and hand you the two or three findings that earn a comment, with the reasoning attached so you can check them in seconds.

npx skills add priyank766/OpenSource-SKILL
INSTALL ONE SKILL
ONE MEDIUM PULL REQUEST 12raw findings 3reach you The other nine were correct too. They just were not worth your interrupt.
01 · Why

Your attention
is the scarce
resource

Reviewing a pull request well takes two things that are hard to do quickly. Reading the change carefully enough to find what actually breaks — and then deciding which of those findings is worth the author's time. The first is slow. The second is the part everyone skips when the first ran long.

These skills carry both. They spend the time on the reading — reading order, four passes over the diff, sweeping the whole repo for anything the change touched — then score every finding and keep only the ones that clear the bar. What reaches you is short, specific, and already checked against the test suite.

Does the reading

Reading order, four passes over the diff, blast-radius sweeps, and a reproduction before anything is claimed.

Keeps the few

A weighted 10-point score. Below the floor it never reaches you. Related findings merge into one comment.

Reads like a colleague

Concrete claim, minimal patch, one line of why — so you can approve or dismiss it at a glance.

02 · The suite

Eight lenses,
one procedure

Every skill is self-contained. Install one and it works on its own.

FLOOR is the minimum score a finding must reach to be worth posting. 7.5 where a miss is asymmetric and cannot be un-shipped. 8.5 where noise is expensive and the stakes are low — there you cut harder, not softer.

03 · The filter

Scoring is subtractive

Every candidate finding is scored on correctness, architectural impact, urgency, and how much it costs the author to act on. The floor decides what reaches you.

ScoreBandAction
9.0–10Subtle bug, race, leak, edge-case crash, security flawMust post, with a patch
8.0–8.9Type safety, test isolation, structural polishPost as a suggestion
6.0–7.9Cosmetic refactor, micro-optimization, preferenceDo not post on the PR
< 6.0Generic lint, praise, hallucinationReject silently

Three separate 7.5 nits about the same pattern do not add up to a 9.0. They add up to one 8.5 about the pattern, posted at the first occurrence. Hard budget: one comment per ~200 changed lines, five max.

Typical AI output 6.0

Good job on the error handling! However, catching a broad exception here can be restrictive. Consider narrowing the exception type for better error handling.

Polite, plausible, unactionable. Names no failure mode, so the author cannot tell whether it matters.

What this produces 9.5

This now catches ConnectionError too, so a network blip reaches the caller as a ConfigParseError and the retry at loader.py:64 never fires:
except ParseError as exc:
    raise ConfigParseError(path) from exc
WDYT?

Same observation underneath. One is a notification; the other names the failure and carries the patch. Illustrative — not a real project.

04 · In use

You ask in plain English

Descriptions carry explicit trigger conditions, so the agent routes to the right skill without you naming it.

review — bash
~/kueue $ review PR #482
pr-review · ran the suite, 251 pass on the branch, 242 on main
1 blocking · exhausted retries return None, caller crashes · 9.8
2 posted · 5 rejected under the floor · 4 merged into one
~/kueue $ security pass on the auth changes
security-review · floor 7.5 · traced 6 sources to 11 sinks
1 finding not posted publicly · routed to SECURITY.md disclosure
~/kueue $ this PR bumps 40 packages, anything worrying?
dependency-review · lockfile diff, not the manifest
2 install-time scripts in newly added transitive packages
1 mutable action tag in .github/workflows · 8.5
~/kueue $ anything here break downstream users?
api-compat-review · default changed on a public helper
behavioural break, unchanged signature · no CHANGELOG entry · 9.5
~/kueue $

Illustrative session. Numbers are not from a real repository.

05 · Install

Any agent,
or none

Plain Markdown with standards-compliant frontmatter. Any agent that reads a SKILL.md works — and a human can read them directly.

Everything:

npx skills add priyank766/OpenSource-SKILL

A single skill — nothing it does not need is copied:

npx skills add priyank766/OpenSource-SKILL --skill security-review

Also targets Cursor, Gemini, Antigravity, opencode, Copilot, and Cowork — see the install matrix.