quality plugins

by Ivan-WG

Prices every output as a debit and asks how much. Five surfaces, one bar each: code, tests, comments, files, and actions — a tool call, subagent spawn or verify pass whose result cannot change the next step is a paid no-op, and nothing else here prices that. Only four things buy more than the minimum: blast radius, an observed defect, a stated criterion, the user asking; the excess must name which, where it happens. Minimum means risk coverage, not count — never delete a test to hit a ratio.

by dunnock

Conduct Pareto Analysis (80/20 Rule) to identify the vital few causes driving the majority of problems. Guides data collection, category definition, chart creation, cumulative percentage calculation, and prioritization. Generates professional Pareto charts (SVG) and HTML reports with quality scoring. Use when prioritizing defects, complaints, failures, or improvement opportunities; when user mentions "Pareto", "80/20 rule", "vital few", "trivial many", "prioritization", or needs to identify which factors contribute most to a problem.

by Kyle Tabor

Pre-mortem analysis skill — define success/failure criteria before execution. Forces agents to articulate what bad output looks like before starting multi-step work. Based on Gary Klein's prospective hindsight research.

by Nery Samuel Murillo

Format files edited by Codex with Prettier after each change.

by Ivan-WG

Meta-bundle: the code-quality disciplines that advise rather than enforce — approach deliberation and design patterns, OWASP security review and threat modeling, systematic debugging, resilience (failure modes, error handling, concurrency, observability, measure-first performance), dependency hygiene and stack inventory (stack-scan), and test design and coverage review. Accessibility auditing lives in ui-ux (frontend-suite, craft-suite). Split out of quality-suite so a project can take the enforcing half without the advisory half's always-on context; the membership rule is prose, not script-checked. Uninstall cleanly with /quality-principles-suite:uninstall. 200k-window installs: see the README's context-window requirement first.

by Ivan-WG

Meta-bundle: the code-quality plugins that carry an enforcement mechanism — review and code smells with the single-pass fan-in, architecture's evidence-at-claim Stop gate, candor's citation-and-reversal Stop gate, comment discipline (warn-on-write plus a narrow pre-write deny), destructive-command blocking, secret-leak prevention, a token-cost warn hook (lean), and file-aware skill auto-routing. The advisory nine live in quality-principles-suite; the membership rule is prose, not script-checked. Uninstall cleanly with /quality-suite:uninstall.

by dunnock

Orchestrate complete Root Cause and Corrective Action (RCCA) investigations using the 8D methodology. Guides team formation (D1) with domain-specific recommendations, problem definition (D2), containment (D3), root cause analysis (D4) with integrated tool selection (5 Whys, Fishbone, Pareto, Kepner-Tregoe, FTA), corrective action (D5-D6), prevention (D7), and closure (D8). Use when conducting RCCA, 8D, root cause analysis, corrective action, failure investigation, nonconformance analysis, quality problems, or customer complaints.

by Steve Harlow

Audit Claude Code artifacts from inside Claude Code: the audit-skill runs the free skillcrossroads CLI to grade skills, subagents, slash commands, MCP configs, and plugins with an evidence-cited file:line scorecard, then walks the ranked fix list until the grade stops improving.

by Steve Harlow

Audit your Claude Code skills from inside Claude Code: installs the audit-skill, which runs the skillcrossroads CLI (the grader for skills, subagents, slash commands, MCP configs, and plugins) to produce an evidence-cited, file-and-line scorecard and walk the ranked fix list until the grade stops improving.

by Kleber Maia

Make your coding agent work like a staff engineer: interview before building, preview before tests, gated lifecycle, plain-language handoffs. Stack-agnostic, zero dependencies.

by Ivan-WG

Chat-message brevity as a shape contract, not a word filter: prose-line budgets per turn kind, a fixed work-done skeleton (verdict → artifacts → max 5 findings → blocker → next), and a named cut list. Levels lite/full/ultra plus wenyan (classical-Chinese novelty) variants, hook-reinforced each turn. Out of scope: code, files written, subagent prompts, and how much work a turn does. Ships /terse:commit, /terse:compress, /terse:check, three compressed-return agents, and an opt-in MCP catalog shrinker.

by Ivan-WG

Five expert reviewers that RUN the project's own static analysis instead of reciting a best-practice checklist: php-expert (phpstan, psalm, phpcs, rector), ts-expert (tsc, eslint, biome, oxlint), react-expert (the hook rules), vue-expert (vue-tsc, eslint-plugin-vue) and ui-expert (stylelint, pa11y, lighthouse-ci, axe). A detect-analyzers.sh mechanism reports which tool is configured, which one CI actually enforces, what the baseline already forgives and which rules are switched off — so the headline finding is the analyzer's blind spot, which no remembered rubric can produce. Read-only; fixes route to the existing workers.

by Kyle Tabor

DEPRECATED (2026-09-16) — hooks removed; plugin is a no-op. Independent verification lives in dev-process. See README.

by Clock Lobster

A disciplined, four-gate bug-fixing methodology for coding agents: Concern (failing repro) -> Cause (root cause + siblings) -> Countermeasure (fix the architecture) -> Check (red-then-green proof + hardening). Reproduce-first, regression-proof, blast-radius-free fixes.

by Tamas Mezei

A suite for authoring Claude Code artifacts: reviewing-claude-skills and reviewing-claude-subagents each review their artifact in two passes — structure first, stopping at a structural verdict when the shape fails (fit-for-purpose leads the subagent gate), otherwise a full sweep ending in a computed verdict that honors the artifact's recorded waivers; improving-claude-skills and improving-claude-subagents loop the matching review autonomously — review → apply every blocking finding → commit → re-review — until the verdict is acceptable, the findings plateau, or the round cap is hit, asking only one question: the intent brief at kickoff. The subagent loop ends instead of converting when a review says the subagent should be a skill.

by daishengda2018

Android code review toolkit v3.1.1 - Refactored architecture with clear responsibility separation

by Carlos Araujo

Code review de PR com veredicto GO/NO-GO por finding sobre o merge e fluxo de backlog para findings diferiveis.

by Renford Nelson

A standalone code review skill for Claude Code — evidence-tiered findings, a required decision model, combined-findings anti-blur rules, and a single-clarifying-question resume contract, independent of any other plugin. Renders via ReportFindings and an optional review-dashboard Artifact; per-file review state persists to a caller-supplied location or stays ephemeral for a single pass.

by wdx9413

Standalone repeatable evaluation and promotion evidence for Agent assets and task outcomes.

by Eduardo Furihata

Skills de construção do Furihata — /solve, /principles, /front, /method e /proto: do problema ao commit local, com QA. Funciona sozinho, em qualquer repositório.

by Cove

Evidence-gated review and bounded repair of a concrete Git change scope. Implicitly use only to assess uncommitted changes, a branch or diff, a local ref range, or a PR for material defects, regressions, test gaps protecting changed behavior, or merge readiness. Do not implicitly use for document or generated-output review, output diagnosis, general skill, plugin, or repository analysis, architecture exploration, or planning-only work.

by OpenAI Codex

Evaluate Codex skills and plugins from chat with a beginner-friendly start command, local-first reports, token budget explanations, and guided benchmarking.

by Owen Ayres

Three-layer quality evaluation framework for agent plugins and skills (Agent Skills / Agent Plugins 1.0) with Elo ranking

by Tamas Mezei

Supplies criteria groups B–G for grading a Markdown prompt that steers Claude — model-specific prompting, general prompting, hallucination guards, output consistency, injection defenses, and prompt-leak defenses. Returns the criteria; the caller scores against them.