An open library of skills for AI agents — ranked by quality, not installs.
skills.sh is great at discovery and popularity. Knackbox is the quality layer on the same open Agent Skills format: mandatory benchmarks, trust digests, honest metrics (no faked scores), and curated packs. See WHY.md · STANDARD.md · why page.
Skills are small folders of instructions that teach a model a job well — commits, reviews, SQL, incidents, writing — for Claude Code, Codex, Cursor, OpenCode, and any runtime that reads SKILL.md.
skills/<category>/<skill-name>/
├── SKILL.md # required: YAML frontmatter + instructions
├── scripts/ # optional executable helpers
├── references/ # optional docs the model loads only when needed
└── assets/ # optional templates, fonts, icons
The SKILL.md frontmatter has two required fields. The description is what makes a model decide to use the skill, so it always says when to trigger, not just what the skill does:
---
name: commit-messages
description: Write clear, conventional git commit messages. Use whenever the
user asks for a commit message, is about to commit changes, or shares a
diff that needs summarizing.
---Knackbox CLI — zero-dependency Node installer (resolves the live catalog, extracts from the repo tarball):
npx knackbox add commit-messages
npx knackbox add code-review security-review --agent cursor
npx knackbox pack developer-essentials
npx knackbox search security
npx knackbox list --sort tier
npx knackbox compare code-review security-review
npx knackbox add commit-messages --agent codex,cursor
npx knackbox doctor
npx knackbox whyDefault destination is ~/.claude/skills/<name>. Use --agent codex|cursor|opencode, --dest DIR, or --force as needed. Package source: packages/cli. Job-based install recipes: RECIPES.md.
Until the package is on the public npm registry, run from a clone:
node packages/cli/bin/knackbox.js add commit-messages
# or: npm run knackbox -- add commit-messagesOpen Skills CLI (skills.sh) — install into Claude Code, Codex, OpenCode, Cursor, or another detected agent:
npx skills add shinzoxD/knackbox --list
npx skills add shinzoxD/knackbox --skill commit-messages -y
npx skills add shinzoxD/knackbox --skill '*' -yAdd -g for a global install or -a claude-code (or -a '*' for all agents). Knackbox is fully compatible with the open Agent Skills ecosystem; see ECOSYSTEM.md for how skills.sh works and how to combine third-party skills with this library.
Dependency-free fallback — install directly into Claude Code with the repository script:
curl -fsSL https://raw.githubusercontent.com/shinzoxD/knackbox/main/install.sh | bash -s commit-messagesUse --dest DIR to select another Agent Skills-compatible directory, or --force to replace an existing copy.
Claude.ai — open Settings → Capabilities → Skills and upload the skill folder (or paste the SKILL.md).
Anything else — any agent runtime that supports the Agent Skills format can load these folders as-is.
Curated packs provide explicit multi-skill installation commands for common workflows:
| Pack | Skills |
|---|---|
| Developer Essentials | Design, debug, test, review, secure, CI/CD, k8s, ship |
| Quality & Accessibility | Code review, security, a11y, i18n, performance, triage |
| Platform & Infra | Terraform, k8s, containers, CI/CD, load tests, GraphQL |
| Data Workbench | Cleaning, analysis, charts, metrics, experiments, SQL |
| Research Rigor | Fact checks, summaries, literature, competitors, surveys |
| Clear Writing | Email, blogs, launches, social, internal & API comms |
| Project Leadership | Decisions, RFCs, OKRs, onboarding, status, retros |
| Ops & Reliability | Runbooks, postmortems, k8s, load tests, observability |
| Career & Hiring | Interview prep, resumes, decision memos, professional email |
| Customer Facing | Support macros, privacy, incident updates, announcements |
| Privacy & Trust | Privacy requests, support judgment, security, agent hardening |
| Team Operating System | OKRs, agendas, onboarding, status, retros, delegation |
Job-based install sets: RECIPES.md.
Browse commands at knackbox.pages.dev/packs or consume packs.json directly.
Want to help grow the library? See COMMUNITY.md for good first contributions, skill ideas, and how to request missing workflows. New skills and improvements are welcome — CONTRIBUTING.md has the quality bar.
Skills are ranked the way Artificial Analysis ranks models — independent measurement columns plus a composite score, with a coarse tier for quick filtering: ⭐ core (maintainer-curated, suite-covered, and maintained), ✅ verified (suite-covered and manually validated until measured thresholds are published), 🧪 community (passes CI validation). Every skill ships a benchmark suite; unmeasured score fields remain visibly empty rather than being estimated. Full methodology, formulas, and how to get your skill Verified: METRICS.md.
The table below is regenerated automatically by CI on every merge — do not edit it by hand.
| Tier | Skill | Category | Context (tok) | Description |
|---|---|---|---|---|
| ⭐ core | code-review | coding | 572 | Review code changes for bugs, security issues, and maintainability with severity-tagged, actionable findings.… |
| ⭐ core | commit-messages | coding | 496 | Write clear, conventional git commit messages from diffs or descriptions of changes. Use whenever the user as… |
| ⭐ core | meeting-notes | documents | 536 | Turn raw meeting transcripts, chat logs, or rough notes into clean, structured meeting notes with decisions a… |
| ✅ verified | debugging | coding | 541 | Diagnose bugs systematically instead of guessing — form hypotheses, isolate the fault, and verify the fix. Us… |
| 🧪 community | accessibility-review | coding | 441 | Review UI and content for accessibility issues against WCAG-minded practices. Use whenever the user asks for… |
| 🧪 community | api-changelog | coding | 254 | Write consumer-facing API changelogs and migration notes for breaking and additive changes. Use whenever the… |
| 🧪 community | api-design | coding | 674 | Design and review HTTP, REST, GraphQL, and internal service APIs with clear contracts, errors, pagination, se… |
| 🧪 community | caching-strategies | coding | 286 | Design caching layers with correctness, TTLs, invalidation, and stampede control. Use whenever the user asks… |
| 🧪 community | chaos-experiment-design | coding | 302 | Plan chaos and resilience experiments safely with hypotheses, blast radius limits, and abort criteria. Use wh… |
| 🧪 community | ci-cd-pipelines | coding | 340 | Design and review CI/CD pipelines for speed, safety, and clear failure signals. Use whenever the user edits G… |
| 🧪 community | cli-design | coding | 244 | Design developer CLIs with clear UX, exit codes, flags, and help text. Use whenever the user builds a command… |
| 🧪 community | code-comments | coding | 317 | Write and improve code comments and docstrings that explain why, not noise. Use whenever the user asks for be… |
| 🧪 community | container-review | coding | 320 | Review Dockerfiles and container configs for size, security, and operability. Use whenever the user shares a… |
| 🧪 community | database-migrations | coding | 377 | Plan and review database schema migrations for safety and rollback. Use whenever the user writes migrations,… |
| 🧪 community | dependency-upgrades | coding | 648 | Plan and execute dependency upgrades with risk assessment, changelog review, and verification. Use whenever t… |
| 🧪 community | design-system-contribution | coding | 328 | Guide contributions to a design system — component API, accessibility, tokens, and docs. Use whenever the use… |
| 🧪 community | error-handling | coding | 365 | Design robust error handling, retries, and failure modes for application code. Use whenever the user asks abo… |
| 🧪 community | feature-flags | coding | 321 | Design feature flag rollout plans with targeting, kill switches, and cleanup. Use whenever the user asks abou… |
| 🧪 community | git-conflict-resolution | coding | 374 | Resolve git merge and rebase conflicts carefully while preserving intent. Use whenever the user has merge con… |
| 🧪 community | graphql-schema-design | coding | 354 | Design and review GraphQL schemas for authz, pagination, N+1 risk, nullability, and evolution. Use whenever t… |
| 🧪 community | helm-chart-scaffolding | coding | 332 | Scaffold and review Helm charts with safe defaults for values, templates, and hooks. Use whenever the user as… |
| 🧪 community | i18n-review | coding | 434 | Review product UI and copy for internationalization readiness — concatenation, pluralization, locale formats,… |
| 🧪 community | kubernetes-review | coding | 380 | Review Kubernetes manifests and Helm values for security, reliability, and operability. Use whenever the user… |
| 🧪 community | load-test-planning | coding | 355 | Plan and interpret load tests with clear goals, scenarios, and pass/fail criteria. Use whenever the user asks… |
| 🧪 community | logging-observability | coding | 382 | Design logging, metrics, and tracing that operators can use under pressure. Use whenever the user asks about… |
| 🧪 community | migration-playbooks | coding | 258 | Plan large technical migrations with strangler patterns, dual-run, and rollback. Use whenever the user migrat… |
| 🧪 community | observability-alerts | coding | 259 | Design actionable alerts that page humans only for real pain. Use whenever the user writes alert rules, fixes… |
| 🧪 community | openapi-spec-design | coding | 275 | Design and review OpenAPI specifications for clarity, consistency, and evolution. Use whenever the user write… |
| 🧪 community | performance-review | coding | 641 | Diagnose application performance bottlenecks and propose measured fixes. Use whenever the user mentions slow… |
| 🧪 community | pr-descriptions | coding | 914 | Write pull request titles and descriptions that reviewers can act on fast. Use whenever the user asks for a P… |
| 🧪 community | prompt-injection-hardening | coding | 559 | Threat-model AI features against prompt injection, tool abuse, and data exfiltration, then propose mitigation… |
| 🧪 community | rate-limiting | coding | 277 | Design rate limiting and throttling for APIs and gateways with fair use and abuse resistance. Use whenever th… |
| 🧪 community | refactor-planning | coding | 630 | Plan safe refactors that preserve behavior — seams, steps, tests, and rollback. Use whenever the user asks ho… |
| 🧪 community | secrets-management | coding | 347 | Handle secrets safely in apps and infrastructure without committing credentials. Use whenever the user asks a… |
| 🧪 community | security-review | coding | 1095 | Review code and designs for security vulnerabilities — injection, authn/authz, secrets, crypto, SSRF, and dat… |
| 🧪 community | sql-query-review | coding | 948 | Review and improve SQL queries for correctness, performance, safety, and maintainability. Use whenever the us… |
| 🧪 community | terraform-review | coding | 440 | Review Terraform and OpenTofu configuration and plans for blast radius, state risk, IAM overreach, and safer… |
| 🧪 community | test-generation | coding | 616 | Design and write focused automated tests for new code, bug fixes, and risky behavior. Use whenever the user a… |
| 🧪 community | websocket-design | coding | 286 | Design WebSocket and realtime channels with auth, backpressure, and reconnect semantics. Use whenever the use… |
| 🧪 community | ab-test-analysis | data | 377 | Analyze A/B and experiment results with correct metrics, uncertainty, and caveats. Use whenever the user shar… |
| 🧪 community | anomaly-detection | data | 257 | Investigate metric and log anomalies with structured hypotheses and next checks. Use whenever the user sees a… |
| 🧪 community | csv-analysis | data | 961 | Analyze CSV, TSV, and tabular data files — profile the data, answer questions with verifiable numbers, and fl… |
| 🧪 community | data-cleaning | data | 540 | Profile and clean messy tabular datasets with reproducible transformations and explicit handling of types, mi… |
| 🧪 community | data-contract-design | data | 308 | Define producer-consumer data contracts with schema evolution, compatibility, and SLAs. Use whenever the user… |
| 🧪 community | data-visualization | data | 579 | Choose and produce clear, honest charts from tabular or summarized data. Use whenever the user asks for a cha… |
| 🧪 community | etl-pipeline-design | data | 300 | Design ETL and ELT pipelines with clear contracts, quality checks, and failure modes. Use whenever the user p… |
| 🧪 community | metrics-definitions | data | 378 | Define product and business metrics with precise numerators, denominators, and edge cases. Use whenever the u… |
| 🧪 community | sql-analytics | data | 294 | Write analytical SQL for warehouses and BI questions with correct grain and caveats. Use whenever the user as… |
| 🧪 community | architecture-decision-records | documents | 332 | Write concise Architecture Decision Records that capture context, decision, and consequences. Use whenever th… |
| 🧪 community | incident-postmortems | documents | 556 | Write blameless incident postmortems with timeline, impact, root cause, and action items. Use whenever the us… |
| 🧪 community | onboarding-guides | documents | 292 | Write new-hire and contributor onboarding guides that get people productive fast. Use whenever the user asks… |
| 🧪 community | policy-writing | documents | 208 | Write clear internal policies people can follow under time pressure. Use whenever the user asks for a company… |
| 🧪 community | project-proposals | documents | 477 | Turn a rough initiative into a decision-ready project proposal with outcomes, scope, plan, costs, risks, and… |
| 🧪 community | release-notes | documents | 724 | Write clear user-facing and developer release notes from commits, PRs, and changelogs. Use whenever the user… |
| 🧪 community | resume-builder | documents | 526 | Build or rewrite resumes and tailor them to a specific job description, with achievement-focused bullets that… |
| 🧪 community | rfc-design-docs | documents | 588 | Write RFCs and technical design docs that force clear decisions. Use whenever the user asks for an RFC, desig… |
| 🧪 community | rfp-response | documents | 279 | Structure RFP and RFI responses that map requirements to evidence. Use whenever the user answers a vendor que… |
| 🧪 community | runbook-writing | documents | 356 | Write operational runbooks that on-call engineers can follow under stress. Use whenever the user asks for a r… |
| 🧪 community | sla-slo-writing | documents | 255 | Write SLAs and SLOs that operations can measure and support can explain. Use whenever the user drafts service… |
| 🧪 community | status-reports | documents | 245 | Write clear project and program status reports for stakeholders. Use whenever the user asks for a weekly stat… |
| 🧪 community | technical-documentation | documents | 504 | Write and improve technical documentation, tutorials, runbooks, and reference pages that users can follow suc… |
| 🧪 community | vendor-evaluation | documents | 252 | Run structured vendor evaluations and scorecards for build-vs-buy decisions. Use whenever the user compares v… |
| 🧪 community | 1-1-agendas | productivity | 224 | Prepare effective manager and IC 1:1 agendas and notes. Use whenever the user asks for a one-on-one agenda, 1… |
| 🧪 community | decision-memos | productivity | 460 | Write concise decision memos that compare options, expose trade-offs, and make a clear evidence-based recomme… |
| 🧪 community | delegation-briefs | productivity | 248 | Write tight delegation briefs so others can execute without thrash. Use whenever the user wants to hand off w… |
| 🧪 community | email-drafts | productivity | 478 | Draft and rewrite professional emails that get replies — clear ask, right tone, minimal length. Use whenever… |
| 🧪 community | hiring-scorecards | productivity | 277 | Build interview scorecards and debrief structures that reduce bias and noise. Use whenever the user designs h… |
| 🧪 community | interview-prep | productivity | 354 | Prepare for technical and behavioral interviews with structured practice. Use whenever the user asks for inte… |
| 🧪 community | issue-triage | productivity | 405 | Triage bugs and tickets with severity, repro steps, and next actions. Use whenever the user asks to triage is… |
| 🧪 community | meeting-agendas | productivity | 320 | Build focused meeting agendas with outcomes, timeboxes, and prep. Use whenever the user asks for a meeting ag… |
| 🧪 community | okr-drafting | productivity | 303 | Draft OKRs and goal cascades that are measurable and non-vanity. Use whenever the user asks for OKRs, quarter… |
| 🧪 community | oss-maintainer-triage | productivity | 326 | Triage open-source issues and PRs with labels, reproduction asks, and maintainer-friendly replies. Use whenev… |
| 🧪 community | privacy-request-playbook | productivity | 403 | Handle DSAR, GDPR, and CCPA-style privacy requests with verification, scoping, and careful response templates… |
| 🧪 community | prompt-improver | productivity | 540 | Rewrite and strengthen prompts for AI models — diagnose what's vague, add structure, and return a copy-paste-… |
| 🧪 community | retro-facilitation | productivity | 507 | Run and write software retros with useful insights and follow-through. Use whenever the user asks for a retro… |
| 🧪 community | sprint-planning | productivity | 281 | Run sprint or iteration planning with capacity, goals, and clear commit scope. Use whenever the user asks for… |
| 🧪 community | standup-updates | productivity | 367 | Turn raw work notes into crisp standup or status updates. Use whenever the user asks for a standup, daily sta… |
| 🧪 community | support-macros | productivity | 381 | Draft customer support replies and macros with correct tone, escalation judgment, and no over-promising. Use… |
| 🧪 community | task-planning | productivity | 498 | Convert ambiguous work into an executable plan with outcomes, dependencies, milestones, owners, and verificat… |
| 🧪 community | timebox-planning | productivity | 253 | Plan deep work and timeboxes when everything feels urgent. Use whenever the user asks how to schedule focus t… |
| 🧪 community | competitor-briefs | research | 532 | Research competitors into structured comparison briefs with evidence and gaps. Use whenever the user asks for… |
| 🧪 community | customer-discovery | research | 260 | Plan and run customer discovery interviews that test assumptions. Use whenever the user is validating a start… |
| 🧪 community | fact-checking | research | 549 | Verify factual claims against current, authoritative evidence and return calibrated verdicts with citations.… |
| 🧪 community | literature-reviews | research | 548 | Plan and synthesize a literature review across papers and reports, organizing evidence by research question r… |
| 🧪 community | market-sizing | research | 249 | Build transparent TAM SAM SOM style market sizing with explicit assumptions. Use whenever the user asks for m… |
| 🧪 community | research-summaries | research | 496 | Summarize papers, articles, and reports into structured research briefs that separate claims from evidence. U… |
| 🧪 community | survey-analysis | research | 266 | Analyze survey results into themes, segments, and decision-ready findings. Use whenever the user shares surve… |
| 🧪 community | user-interview-synthesis | research | 358 | Synthesize user interviews into themes, insights, and opportunities. Use whenever the user pastes interview n… |
| 🧪 community | blog-posts | writing | 856 | Write blog posts and articles with a strong hook, clear structure, and a consistent voice. Use whenever the u… |
| 🧪 community | case-studies | writing | 234 | Write customer case studies with problem, solution, proof, and quote discipline. Use whenever the user asks f… |
| 🧪 community | copy-editing | writing | 522 | Edit prose for clarity, correctness, flow, consistency, and concision while preserving the author's meaning a… |
| 🧪 community | documentation-migration | writing | 290 | Plan and execute docs migrations between tools and information architectures. Use whenever the user moves doc… |
| 🧪 community | incident-comms | writing | 377 | Write customer and stakeholder communications during incidents. Use whenever the user needs a status page upd… |
| 🧪 community | internal-comms | writing | 207 | Write clear internal company communications — all-hands notes, policy updates, and reorg announcements. Use w… |
| 🧪 community | landing-page-copy | writing | 503 | Write and improve landing-page copy with a clear offer, credible proof, scannable structure, and focused call… |
| 🧪 community | newsletter-writing | writing | 161 | Write email newsletters with a clear point, scannable structure, and one CTA. Use whenever the user asks for… |
| 🧪 community | press-release | writing | 171 | Write press-style announcements with a lead, facts, and quotes without hype fluff. Use whenever the user need… |
| 🧪 community | product-announcements | writing | 327 | Write product launch and feature announcements for users and stakeholders. Use whenever the user asks for a l… |
| 🧪 community | social-posts | writing | 291 | Write clear social posts and short threads for LinkedIn, X, and similar channels. Use whenever the user asks… |
The same data is available as machine-readable catalog.json, which also powers the project website. Each entry includes its source URL, Apache-2.0 license, compatibility notes, declared network/filesystem/execution permissions, executable-code profile, and a deterministic SHA-256 digest of the complete skill package.
Live catalog: knackbox.pages.dev — browse the skill leaderboard, starter packs, and per-skill pages with install commands.
The static site in site/ is built from catalog.json plus the raw repository markdown files. It ships with SEO essentials: canonical URLs, Open Graph / Twitter cards, JSON-LD, sitemap-index.xml, robots.txt, and llms.txt for AI crawlers. Machine-readable catalog.json is published at the site root.
Requires Node.js 22.12 or newer.
Local development:
cd site
npm i
npm run devProduction build:
cd site
npm run buildThe build copies catalog.json / packs.json into public/, emits static files to site/dist, generates the sitemap, and runs Pagefind as a postbuild search indexer. Deploy site/dist to any static host; the included GitHub Actions workflow is wired for Cloudflare Pages.
New skills are very welcome — the whole point of this repo is to grow. Read CONTRIBUTING.md for the folder rules and quality bar, copy template/SKILL.md to get started, and run the validator before opening a PR:
python scripts/validate.pyLooking for ideas? Check the skill request issues.
Questions and early proposals belong in GitHub Discussions. See SUPPORT.md for where to ask for help, GOVERNANCE.md for how decisions and tiers are managed, and EVALUATION.md for the reproducible benchmark protocol.
Skills may contain executable scripts. Every script is reviewed line-by-line before merge, but you should still read what you install — see SECURITY.md.
Apache-2.0 — see LICENSE. Contributions are accepted under the same license via DCO sign-off.