Run any Skill in Manus with one click

code-review

Stars7

Forks1

UpdatedApril 9, 2026 at 02:22

Guidelines for AI code review — finding real bugs, not style nits. TRIGGER when reviewing code, diffs, PRs, or implementations.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

paulgp

paulgp/replication-package-db

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software Quality Assurance Analysts and TestersComputer and Mathematical Occupations·SOC 15-1253

SKILL.md

readonly

name	code-review
description	Guidelines for AI code review — finding real bugs, not style nits. TRIGGER when reviewing code, diffs, PRs, or implementations.

Code Review

Find bugs that would break production. Not improvements. Not style.

A Finding Must Be

Introduced in this change — not pre-existing
Provably impactful — name the scenario/input that triggers it. Speculation is not a finding.
Actionable — a discrete fix exists

Severity

P0 — Blocks merge. Breaks production, loses data, security hole.
P1 — Should fix before merge. Realistic conditions trigger it.
P2 — Worth fixing eventually. Unlikely in practice.
P3 — Nit. Do not report.

Don't inflate. P2 called P0 erodes trust.

Verdict

Every review ends with exactly one:

LGTM — no P0/P1. Ship it.
FAIL — P0/P1 found. Must fix.

LGTM = short. Don't pad with praise.

Process

Read plan.md if it exists
Read changed files (use read tool, don't guess from context)
Run tests — failure is P0
If shared state changed, run race detector

Output

## Findings

### [P0] Title
**File**: path.go:42-48
**Issue**: what's wrong and what triggers it
**Fix**: what to change

## Verdict
LGTM / FAIL

More from this repository

same repository

forge-orchestration

paulgp/replication-package-db

Lead multi-agent workflows using forge tools (dispatch, debate, models). TRIGGER when asked to ship issues, debate plans, review code with multiple agents, or use /ship /debate /plan /review commands.

2026-04-097

forgejo-actions

paulgp/replication-package-db

Write and debug Forgejo Actions workflows. TRIGGER when creating/fixing CI jobs, retrieving job logs, diagnosing build failures, or configuring runners.

2026-04-097

forgejo

paulgp/replication-package-db

Manages repositories, issues, PRs, releases, and CI on Forgejo via REST API. TRIGGER when needing to create/edit issues, open PRs, view CI status, manage repos, or check releases.

2026-04-097

model-selection

paulgp/replication-package-db

Discover available LLM models with live benchmarks, pricing, and throughput, while enforcing recency and 429-sensitivity for dispatch.

2026-04-097

name	code-review
description	Guidelines for AI code review — finding real bugs, not style nits. TRIGGER when reviewing code, diffs, PRs, or implementations.

Code Review

Find bugs that would break production. Not improvements. Not style.

A Finding Must Be

Introduced in this change — not pre-existing
Provably impactful — name the scenario/input that triggers it. Speculation is not a finding.
Actionable — a discrete fix exists

Severity

P0 — Blocks merge. Breaks production, loses data, security hole.
P1 — Should fix before merge. Realistic conditions trigger it.
P2 — Worth fixing eventually. Unlikely in practice.
P3 — Nit. Do not report.

Don't inflate. P2 called P0 erodes trust.

Verdict

Every review ends with exactly one:

LGTM — no P0/P1. Ship it.
FAIL — P0/P1 found. Must fix.

LGTM = short. Don't pad with praise.

Process

Read plan.md if it exists
Read changed files (use read tool, don't guess from context)
Run tests — failure is P0
If shared state changed, run race detector

Output

## Findings

### [P0] Title
**File**: path.go:42-48
**Issue**: what's wrong and what triggers it
**Fix**: what to change

## Verdict
LGTM / FAIL