agentsclimarketplace
Sign in

Bug review

Skill athola/claude-night-market/plugins/pensive/skills/bug-review

Hunts bugs with evidence trails. Use when investigating unexpected behavior or before merging code with potential hidden defects.From its SKILL.md

Install
npx -y skills add athola/claude-night-market --skill bug-review

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

7.1 KB, ~1.6k tokens by cl100k_base, as published. Nobody here has run it

Table of Contents

Bug Review Workflow

Systematic bug identification and fixing with language-specific expertise.

Quick Start

/bug-review

Verification: Run the command with --help flag to verify availability.

When To Use

  • Reviewing code for potential bugs
  • After receiving bug reports
  • Before major releases
  • During security audits
  • Investigating production issues

When NOT To Use

  • Test coverage audit - use test-review instead

Required TodoWrite Items

  1. bug-review:language-detected
  2. bug-review:repro-plan
  3. bug-review:defects-documented
  4. bug-review:fixes-prepared
  5. bug-review:verification-plan
  6. bug-review:findings-verified

Progressive Loading

Load additional context as needed:

  • Language Detection: @include modules/language-detection.md - Manifest heuristics, expertise framing, version constraints
  • Defect Documentation: @include modules/defect-documentation.md - Severity classification, root cause analysis, static analyzers
  • Fix Preparation: @include modules/fix-preparation.md - Minimal patches, idiomatic patterns, test coverage

Workflow

Step 1: Detect Languages (bug-review:language-detected)

Identify dominant languages using manifest files (Cargo.toml → Rust, package.json → Node, etc.).

State expertise persona appropriate for the language ecosystem.

Note version constraints (MSRV, Python versions, Node engines).

Progressive: Load modules/language-detection.md for detailed manifest heuristics.

Step 2: Plan Reproduction (bug-review:repro-plan)

Identify reproduction methods:

  • Unit/integration test suites
  • Fuzzing tools
  • Manual reproduction commands

Document exact commands:

cargo test -p core
pytest tests/test_api.py
npm test -- pkg

Verification: Run pytest -v tests/test_api.py to verify.

Capture blockers and propose mocks when dependencies unavailable.

Step 3: Document Defects (bug-review:defects-documented)

Review code line-by-line, logging each bug with:

  • File:line reference: Precise location
  • Severity: Critical, High, Medium, Low
  • Root cause: Logic error, API misuse, concurrency, resource leak
  • Impact: What breaks and how

Run static analyzers (cargo clippy, ruff check, golangci-lint, eslint).

Use imbue:proof-of-work for reproducible capture.

Progressive: Load modules/defect-documentation.md for classification details and analyzer commands.

Step 4: Prepare Fixes (bug-review:fixes-prepared)

Draft minimal, idiomatic patches using language best practices:

  • Guard clauses (Rust: pattern matching, Python: early returns)
  • Resource cleanup (Go: defer, Python: context managers)
  • Error propagation (Rust: ?, Go: wrapped errors)

Create tests following Red → Green pattern:

  1. Write failing test
  2. Apply minimal fix
  3. Verify test passes

Progressive: Load modules/fix-preparation.md for language-specific patterns and test strategies.

Step 5: Verification Plan (bug-review:verification-plan)

Execute reproduction steps with fixes applied.

Capture evidence:

  • Test output logs
  • Benchmark comparisons
  • Coverage reports

Document remaining risks using imbue:diff-analysis/modules/risk-assessment-framework.

Assign owners and deadlines for follow-up items.

Step 6: Verify Findings Are Grounded (bug-review:findings-verified)

Every defect must cite a real file:line and a verbatim Anchor. Write findings to .review/findings.json and confirm each citation resolves:

python plugins/imbue/scripts/citation_verifier.py \
  --findings .review/findings.json --repo-root .

Drop or label UNVERIFIED any defect the verifier fails (exit 1); only verified defects enter the report. See Skill(imbue:review-core) Step 5 for the protocol and Skill(imbue:structured-output) for the schema.

Defect Classification (Condensed)

Severity: Critical (crash/data loss) → High (broken features) → Medium (degraded UX) → Low (edge cases)

Root Causes: Logic errors | API misuse | Concurrency issues | Resource leaks | Validation gaps

Output Format

## Summary
[Brief scope description]

## Defects Found
### [D1] file.rs:142 - Title
- Severity: High
- Anchor: `verbatim source text at file.rs:142`
- Root Cause: Logic error
- Impact: Data corruption possible
- Fix: [description]

## Proposed Fixes
### Fix for D1
[code diff with explanation]

## Test Updates
[new/updated tests with Red → Green verification]

## Evidence
- Commands executed
- Logs and outputs
- External references

Verification: Run pytest -v to verify tests pass.

Best Practices

  1. Evidence-based: Every finding has file:line reference
  2. Reproducible: Clear steps to reproduce each bug
  3. Minimal fixes: Smallest change that fixes the issue
  4. Test coverage: Every fix has corresponding test
  5. Risk awareness: Document remaining risks with severity scoring

Exit Criteria

  • All defects documented with precise references
  • Every defect carries a file:line + verbatim Anchor, and citation_verifier.py confirmed all citations (exit 0) or unverified defects were dropped or labeled UNVERIFIED
  • Fixes prepared with test coverage verified
  • Verification plan includes commands and expected outputs
  • Remaining risks assessed and owners assigned

What ships with it: 3 files

8.2 KB alongside SKILL.md

modules/

Gives 0 of the 12 instructions most review quality skills give in ~1.6k tokens

Counted across 1,048 of the 1,783 authors here whose files we hold, read 2026-08-07

  • Ask questions one at a timein 81 of 1048, across 64 files
  • Provide a recommended answer for each questionin 73 of 1048, across 50 files
  • Explore the codebase instead of asking answerable questionsin 66 of 1048, across 42 files
  • Resolve dependencies between decisions one-by-onein 42 of 1048, across 17 files
  • Interview the user relentlessly about the planin 38 of 1048, across 13 files
  • Order findings by severityin 31 of 1048
  • Resolve each branch of the decision treein 27 of 1048, across 5 files
  • Run a grilling sessionin 26 of 1048, across 5 files
  • Update CONTEXT.md immediately when a term is resolvedin 26 of 1048, across 11 files
  • Propose precise canonical terms for vague languagein 25 of 1048, across 7 files
  • Create documentation files lazilyin 24 of 1048, across 5 files
  • Assign severity to every findingin 24 of 1048

Said here and by no other author read

  • identify dominant languages using manifest files
  • document exact reproduction commands
  • log each bug with a file and line reference
  • draft minimal patches using best practices
  • create tests using the red-green pattern
  • write findings to the findings file

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,871. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.