{ "source": "https://skillproof.dev", "license": "CC BY 4.0 — attribution to skillproof.dev required", "verdicts": { "pass": "installed, triggered, and beat the no-skill baseline on a real task", "setup": "works, but needs a manual step first (named in testNotes)", "fails": "scored below the no-skill baseline: broken, or worse than not installing it" }, "tested": 2172, "returned": 100, "skills": [ { "slug": "statistical-analysis", "name": "Statistical Analysis", "url": "https://skillproof.dev/skills/statistical-analysis", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Enforces frame-inspect-check-assumptions-effect-size-APA-report pipeline for hypothesis tests; bundled Shapiro-Wilk/Levene's script actually runs.", "testNotes": "Actually ran the bundled scripts/assumption_checks.py on synthetic two-group data (n=48/52, no pingouin installed) — check_normality_per_group and check_homogeneity_of_variance both executed and returned correct Shapiro-Wilk (p=.90/.76) and Levene's (p=.33) results, which fed a full APA-style report (t(98)=5.07, p<.001, d=1.02) versus a baseline t-test that skipped assumption checks and effect-size CIs entirely.", "testedDate": "2026-07-14", "repo": "https://github.com/K-Dense-AI/scientific-agent-skills", "install": "git clone https://github.com/K-Dense-AI/scientific-agent-skills\ncd scientific-agent-skills\nmkdir -p ~/.claude/skills\ncp -r skills/statistical-analysis ~/.claude/skills/statistical-analysis", "tags": [ "statistics", "hypothesis-testing", "research", "apa-reporting", "bayesian" ] }, { "slug": "pinme-auth", "name": "Pinme Auth", "url": "https://skillproof.dev/skills/pinme-auth", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "TypeScript reference for PinMe Worker auth: create/verify/list Identity Platform users.", "testNotes": "An unguided client guessed a generic Bearer-token API with an invented pagination scheme; the skill reproduced the real API-key header, exact response envelope, and the 403-vs-401 distinction the API actually uses.", "testedDate": "2026-07-10", "repo": "https://github.com/glitternetwork/pinme", "install": "git clone https://github.com/glitternetwork/pinme\ncd pinme\nmkdir -p ~/.claude/skills\ncp -r skills/pinme-auth ~/.claude/skills/pinme-auth", "tags": [ "pinme", "authentication", "cloudflare-workers" ] }, { "slug": "code-health-check", "name": "Code Health Check", "url": "https://skillproof.dev/skills/code-health-check", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Read-only, sub-agent-forked code health scanner with a fixed severity-graded report format.", "testNotes": "Ran it against the bundled demo app (a deliberately buggy Express API shipped in the same repo) and it caught all 8 planted issues - the SQL injection, the eval() config parser, both hardcoded secrets, the swallowed error, and even the dead formatCategory() function - with correct CRITICAL/WARNING/INFO severities and file:line citations.", "testedDate": "2026-07-13", "repo": "https://github.com/huangjia2019/claude-code-engineering", "install": "git clone https://github.com/huangjia2019/claude-code-engineering\ncd claude-code-engineering\nmkdir -p ~/.claude/skills\ncp -r 04-Skills/projects/07-skill-fork-demo/.claude/skills/code-health-check ~/.claude/skills/code-health-check", "tags": [ "code-quality", "static-analysis", "security-basics", "sub-agent", "report" ] }, { "slug": "fetch-url-as-markdown", "name": "Fetch URL As Markdown", "url": "https://skillproof.dev/skills/fetch-url-as-markdown", "category": "productivity", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Local trafilatura-first URL-to-Markdown fetcher with a clean exit-code contract for when to fall back to Exa.", "testNotes": "Ran the bundled script live on a real Wikipedia page: it returned clean 44KB markdown with a proper YAML metadata header and zero nav/menu chrome, versus a 30KB naive-strip baseline still full of \"Jump to content Main menu... Donate Create account Log in\" junk; separately confirmed the exit-code contract by hitting a real arXiv PDF and getting the documented exit 4 (no Exa fallback) instead of a false extraction failure.", "testedDate": "2026-07-13", "repo": "https://github.com/CodeAlive-AI/ai-driven-development", "install": "git clone https://github.com/CodeAlive-AI/ai-driven-development\ncd ai-driven-development\nmkdir -p ~/.claude/skills\ncp -r skills/fetch-url-as-markdown ~/.claude/skills/fetch-url-as-markdown", "tags": [ "web-scraping", "markdown", "trafilatura", "content-extraction", "research" ] }, { "slug": "cocoindex", "name": "CocoIndex", "url": "https://skillproof.dev/skills/cocoindex", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Turns vague 'build me a RAG pipeline' requests into working CocoIndex flow code with the real API syntax.", "testNotes": "Asked for a flow that chunks, embeds, and exports docs to Postgres+pgvector: the skill's version used CocoIndex's real decorator/row-iteration API and caught the documented 'assign to item[field], not a local variable' gotcha, while an unguided attempt invented a generic .chunk()/.embed() API that doesn't exist in the library.", "testedDate": "2026-07-14", "repo": "https://github.com/cocoindex-io/cocoindex-claude", "install": "git clone https://github.com/cocoindex-io/cocoindex-claude\ncd cocoindex-claude\nmkdir -p ~/.claude/skills\ncp -r cocoindex ~/.claude/skills/cocoindex", "tags": [ "etl", "vector-database", "embeddings", "knowledge-graph", "python" ] }, { "slug": "refresh-context-map", "name": "Refresh Context Map", "url": "https://skillproof.dev/skills/refresh-context-map", "category": "productivity", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Rebuilds the AIDEV-anchor/AGENTS.md/ADR manifest that powers claude-leverage's PreToolUse context hook.", "testNotes": "Actually cloned the repo and ran the real script live: output matched the SKILL.md's documented format exactly ('Wrote ... (26 files, 45 anchors)') — but a fresh HEAD checkout's own committed manifest was already stale by ~40 days (35->44 anchors), showing the project doesn't reliably run its own opt-in pre-commit hook.", "testedDate": "2026-07-14", "repo": "https://github.com/Filip-Podstavec/claude-leverage", "install": "git clone https://github.com/Filip-Podstavec/claude-leverage\ncd claude-leverage\nmkdir -p ~/.claude/skills\ncp -r skills/refresh-context-map ~/.claude/skills/refresh-context-map", "tags": [ "context-management", "claude-code-hooks", "repo-tooling", "manifest" ] }, { "slug": "sbox", "name": "S&box", "url": "https://skillproof.dev/skills/sbox", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Blocks Unity muscle memory when writing s&box (Source 2 / Facepunch) C# components, Razor UI, and networking.", "testNotes": "Asked for a pickup component in both modes: baseline reflexively wrote MonoBehaviour/transform.position/Destroy(gameObject) Unity code, the skill produced correct Component/[Sync]/[Rpc.Broadcast]/IsProxy s&box code -- exactly the Unity-leakage failure the skill's own README warns about, reproduced live and then fixed.", "testedDate": "2026-07-14", "repo": "https://github.com/gavogavogavo/claude-sbox", "install": "git clone https://github.com/gavogavogavo/claude-sbox\ncd claude-sbox\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/sbox", "tags": [ "sbox", "csharp", "gamedev", "source2", "facepunch" ] }, { "slug": "crawl4ai", "name": "Crawl4AI", "url": "https://skillproof.dev/skills/crawl4ai", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Wraps the Crawl4AI browser scraper for JS-heavy sites, batch crawls, and LLM-free schema extraction.", "testNotes": "Ran the skill's exact recommended command against a live JS-rendered test page: a naive curl+CSS-selector scrape returned zero quotes, while `crwl ... -c wait_until=networkidle` returned all ten quotes with authors and tags cleanly formatted in markdown.", "testedDate": "2026-07-14", "repo": "https://github.com/brettdavies/crawl4ai-skill", "install": "git clone https://github.com/brettdavies/crawl4ai-skill\ncd crawl4ai-skill\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/crawl4ai", "tags": [ "web-scraping", "crawling", "playwright", "data-extraction", "automation" ] }, { "slug": "coding-agent-sessions", "name": "Coding Agent Sessions", "url": "https://skillproof.dev/skills/coding-agent-sessions", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Finds and reconstructs real coding-agent session transcripts across Claude, Codex, OpenCode, and more.", "testNotes": "Ran the bundled finder live against this machine's real ~/.claude/projects and ~/.codex/sessions stores with zero pip installs: it correctly listed actual Claude Code sessions (right cwd, timestamps, model) and actual Codex sessions side by side, then a `read` on one Claude session reconstructed all 10 real subagent transcripts matching the reported subagent_count exactly.", "testedDate": "2026-07-14", "repo": "https://github.com/code-yeongyu/oh-my-openagent", "install": "git clone https://github.com/code-yeongyu/oh-my-openagent\ncd oh-my-openagent\nmkdir -p ~/.claude/skills\ncp -r packages/shared-skills/skills/coding-agent-sessions ~/.claude/skills/coding-agent-sessions", "tags": [ "session-search", "claude-code", "codex", "opencode", "transcripts", "cli" ] }, { "slug": "codex-gpt-image", "name": "Codex GPT Image", "url": "https://skillproof.dev/skills/codex-gpt-image", "category": "design", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Generates gpt-image-2 images via local Codex OAuth, no OPENAI_API_KEY needed.", "testNotes": "Ran the bundled CLI for real, end to end: auth-status found a live Codex session, then `generate` actually hit the Codex Images backend and wrote a genuine 1.85MB PNG of a red bicycle on a white background, matching the prompt precisely, in 44 seconds. No OPENAI_API_KEY was set on the machine, so the baseline (no image tool at all) simply cannot complete the task.", "testedDate": "2026-07-14", "repo": "https://github.com/ningzimu/codex-gpt-image", "install": "git clone https://github.com/ningzimu/codex-gpt-image\ncd codex-gpt-image\nmkdir -p ~/.claude/skills\ncp -r skills/codex-gpt-image ~/.claude/skills/codex-gpt-image", "tags": [ "image-generation", "gpt-image-2", "codex-oauth", "chatgpt", "visual-assets" ] }, { "slug": "national-team-position", "name": "National Team Position", "url": "https://skillproof.dev/skills/national-team-position", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Fetches real Shanghai Exchange ETF-share data to estimate China's 'national team' broad-base ETF positioning.", "testNotes": "Ran the bundled script for real against live Shanghai Exchange data (2026-06-01 to 2026-07-10): it correctly surfaced a concrete, checkable signal -- HS300 ETF shares fell 730 to 429 (亿份, -301 net) since their 6/12 peak while CSI1000/A500/STAR50 held roughly flat to up -- and produced all 8 documented output files (7 PNGs + JSON) with zero errors, matching the README exactly, Chinese labels rendering cleanly.", "testedDate": "2026-07-14", "repo": "https://github.com/Xiaoyuan-Liu/national-team-position", "install": "git clone https://github.com/Xiaoyuan-Liu/national-team-position\ncd national-team-position\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/national-team-position", "tags": [ "china-a-shares", "etf-flows", "akshare", "data-visualization", "central-huijin" ] }, { "slug": "routeros-app-yaml", "name": "RouterOS App YAML", "url": "https://skillproof.dev/skills/routeros-app-yaml", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Writes and validates MikroTik RouterOS /app YAML — a docker-compose lookalike with real differences", "testNotes": "Validated both outputs against the project's own published strict JSON Schema for real: the naive docker-compose-style baseline threw 2 hard schema errors (unexpected 'version', unexpected 'deploy'), the skill-guided YAML validated clean on the first pass.", "testedDate": "2026-07-15", "repo": "https://github.com/tikoci/routeros-skills", "install": "git clone https://github.com/tikoci/routeros-skills\ncd routeros-skills\nmkdir -p ~/.claude/skills\ncp -r routeros-app-yaml ~/.claude/skills/routeros-app-yaml", "tags": [ "routeros", "mikrotik", "yaml", "containers", "docker-compose" ] }, { "slug": "trdizin", "name": "TR Dizin", "url": "https://skillproof.dev/skills/trdizin", "category": "documents", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Stdlib-only CLI for the open TR Dizin (Turkish national academic index) JSON API — search, filters, PDF-to-text", "testNotes": "Ran the real live API end to end — search, field-scoped advanced search, author lookup, the documented ':' -in-query error (exact message and exit code 2), and a full PDF-to-markdown extraction of a real open-access paper via markitdown — every single documented behavior matched observed behavior exactly, including an obscure 'returned floors small limit' quirk. A naive baseline fetch of the human-facing trdizin.gov.tr search page 404'd outright.", "testedDate": "2026-07-15", "repo": "https://github.com/saidsurucu/trdizin-skill", "install": "git clone https://github.com/saidsurucu/trdizin-skill\ncd trdizin-skill\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/trdizin", "tags": [ "academic-search", "turkish", "api", "pdf-extraction", "no-key" ] }, { "slug": "conjoint-design", "name": "Conjoint Design", "url": "https://skillproof.dev/skills/conjoint-design", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "A citation-backed conjoint-experiment methodologist: attribute design, power math, and AMCE/AMIE estimation done right.", "testNotes": "Ran a real A/B on an election-candidate conjoint design: the unguided baseline used Cohen's-d power benchmarks, added a raw interaction dummy for AMIE, and claimed 'voters prefer X' from a significant AMCE - the skill's own text flags all three as documented methodological errors and additionally caught a forced-choice/abstention validity issue specific to the electoral framing.", "testedDate": "2026-07-15", "repo": "https://github.com/scdenney/open-science-skills", "install": "git clone https://github.com/scdenney/open-science-skills\ncd open-science-skills\nmkdir -p ~/.claude/skills\ncp -r plugin/skills/conjoint-design ~/.claude/skills/conjoint-design", "tags": [ "conjoint-analysis", "research-design", "power-analysis", "amce", "social-science" ] }, { "slug": "api-filter", "name": "API Filter", "url": "https://skillproof.dev/skills/api-filter", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Current (4.4+) API Platform QueryParameter filter reference that correctly steers away from soon-removed classes", "testNotes": "Checked its two central claims against the live api-platform/core source on GitHub, not just the skill's word: SearchFilter.php line 136 and OrderFilter.php line 202 both carry the exact `@deprecated since 4.4 ... Removed in 6.0` notices the skill describes -- baseline (memory-based) code used exactly those two soon-to-be-removed classes, the skill's version didn't.", "testedDate": "2026-07-15", "repo": "https://github.com/api-platform/skillset", "install": "git clone https://github.com/api-platform/skillset\ncd skillset\nmkdir -p ~/.claude/skills\ncp -r skills/api-filter ~/.claude/skills/api-filter", "tags": [ "api-platform", "php", "symfony", "laravel", "filters" ] }, { "slug": "br-audit-enrichment", "name": "Better Route Audit Enrichment", "url": "https://skillproof.dev/skills/br-audit-enrichment", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Configures Better Route's audit-log middleware so idempotency keys get hashed and logger failures never break the API response.", "testNotes": "Ran real request objects through the actual better-route PHP classes cloned from Lonsdale201/better-route: a naive baseline audit-logger logged the raw bearer token and raw Idempotency-Key and let a logger outage 500 the whole account-update request, while the skill-guided AuditEnricherMiddleware/AuditMiddleware setup produced a clean SHA-1'd idempotency hash, no token exposure, and swallowed the same logger failure so the request still returned 200 — every claim in the SKILL.md checked out against the live source.", "testedDate": "2026-07-15", "repo": "https://github.com/Lonsdale201/wp-agent-skills", "install": "git clone https://github.com/Lonsdale201/wp-agent-skills.git\nmkdir -p ~/.claude/skills\ncp -r wp-agent-skills/better-route/br-audit-enrichment ~/.claude/skills/br-audit-enrichment", "tags": [ "wordpress", "php", "rest-api", "audit-logging", "better-route" ] }, { "slug": "swmm-anywhere", "name": "SWMM Anywhere", "url": "https://skillproof.dev/skills/swmm-anywhere", "category": "data", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Synthesizes a runnable SWMM stormwater network from OSM+DEM when you have no real pipe data.", "testNotes": "Actually ran it end-to-end on a real 1x1km London Greenwich bbox: it downloaded live OSM/DEM data and produced a real SWMM .inp with exactly 33 outfalls -- matching the SKILL.md's documented spike-test number to the digit -- and independently confirmed the doc's claimed Apple-Silicon pyswmm SIGKILL bug (exit code 137 on direct import) while the skill's own stub avoided it.", "testedDate": "2026-07-15", "repo": "https://github.com/Zhonghao1995/agentic-swmm-workflow", "install": "git clone https://github.com/Zhonghao1995/agentic-swmm-workflow\nmkdir -p ~/.claude/skills\ncp -r agentic-swmm-workflow/skills/swmm-anywhere ~/.claude/skills/swmm-anywhere\npip install aiswmm[anywhere]", "tags": [ "swmm", "stormwater", "gis", "osm", "hydrology" ] }, { "slug": "before-you-refactor", "name": "Before You Refactor", "url": "https://skillproof.dev/skills/before-you-refactor", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "A pre-refactor checklist that catches untested behavior before you 'clean up' code.", "testNotes": "Its checklist caught a real regression a naive refactor introduced: an archived record silently 'claiming' its dedup id so a later active duplicate got dropped — the naive rewrite passed all 5 existing tests anyway and only broke on a case none of them covered; the skill's step to add a characterization test for untested behavior caught it before the edit landed.", "testedDate": "2026-07-16", "repo": "https://github.com/oribarilan/97", "install": "git clone https://github.com/oribarilan/97\nmkdir -p ~/.claude/skills\ncp -r 97/skills/before-you-refactor ~/.claude/skills/before-you-refactor", "tags": [ "refactoring", "testing-discipline", "code-review", "checklist", "regression-prevention" ] }, { "slug": "slide-wright", "name": "Slide Wright", "url": "https://skillproof.dev/skills/slide-wright", "category": "design", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Proposes an invented theme and a real 2-slide preview before it ever builds the full deck.", "testNotes": "Built the same 'Q3 sales results' request two ways and rendered both in a real browser: a direct rewrite produced one plain black-on-white bullet slide, while following the skill produced an actual navigable 2-slide reveal.js deck with an invented dark-navy/amber ledger-grid theme, a stat grid layout, and speaker notes.", "testedDate": "2026-07-16", "repo": "https://github.com/arifszn/slide-wright", "install": "git clone https://github.com/arifszn/slide-wright\nmkdir -p ~/.claude/skills\ncp -r slide-wright/slide-wright ~/.claude/skills/slide-wright", "tags": [ "presentations", "slides", "reveal.js", "design-system", "html" ] }, { "slug": "pyxel", "name": "Pyxel Game Development", "url": "https://skillproof.dev/skills/pyxel", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Builds and verifies Pyxel games using real headless validate/run evidence, not vibes.", "testNotes": "Ran the real pyxel-mcp validate+run tools against two versions of a moving-square game: the naive baseline skipped pyxel.cls() and validate flagged it live, then a headless run proved the bug on screen - screen_grid showed 368 nonzero pixels (a smeared ghost trail) versus exactly 64 (a clean 8x8 sprite) once the skill's cls()-first convention was applied.", "testedDate": "2026-07-16", "repo": "https://github.com/kitao/pyxel-skill", "install": "git clone https://github.com/kitao/pyxel-skill.git ~/.claude/skills/pyxel\nrm -rf ~/.claude/skills/pyxel/.git", "tags": [ "pyxel", "game-dev", "python" ] }, { "slug": "video-vision", "name": "Video Vision", "url": "https://skillproof.dev/skills/video-vision", "category": "productivity", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Local ffmpeg+Whisper pipeline that lets Claude actually watch and transcribe video files", "testNotes": "Ran the real pipeline end to end against a synthetic 6-second test clip: ffmpeg frame extraction and local Whisper transcription both executed for real and returned the exact JSON manifest shape the README documents, down to field names. No network calls beyond the disclosed one-time Whisper model download, no exfiltration, nothing hidden - genuinely gives Claude a video-watching capability it has no other way to get.", "testedDate": "2026-07-16", "repo": "https://github.com/Moh4696/claude-video-vision", "install": "git clone https://github.com/Moh4696/claude-video-vision ~/.claude/skills/video-vision", "tags": [ "video", "transcription", "whisper", "ffmpeg", "multimodal" ] }, { "slug": "compose-app-icon", "name": "Compose App Icon", "url": "https://skillproof.dev/skills/compose-app-icon", "category": "design", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "uv-bundled CLI pair that authors and JSON-Schema-validates Apple Icon Composer .icon packages", "testNotes": "Ran the bundled 25-test pytest suite for real (25/25 passed), then fed the same icon document through create_icon.py two ways: a naive version ignoring the SKILL.md 'Gotchas' section failed schema validation with the exact two errors the doc warns about (UI-label shadow kind, scale-only position missing translation-in-points); the version following the gotchas succeeded, produced a real .icon package, and validate_icon.py reported it VALID with output formatting matching the docs byte-for-byte.", "testedDate": "2026-07-16", "repo": "https://github.com/giginet/apple-icon-composer-skill", "install": "git clone https://github.com/giginet/apple-icon-composer-skill\nmkdir -p ~/.claude/skills\ncp -r apple-icon-composer-skill/plugins/icon-composer/skills/compose-app-icon ~/.claude/skills/compose-app-icon", "tags": [ "apple", "icon-composer", "app-icon", "json-schema", "uv" ] }, { "slug": "patent-writing-chinese", "name": "Patent Writing Chinese", "url": "https://skillproof.dev/skills/patent-writing-chinese", "category": "documents", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Generates CNIPA-style Chinese patent .docx files with real OMML equations and B&W block diagrams.", "testNotes": "Actually ran the bundled example end-to-end (pip install python-docx/lxml/matplotlib, executed examples/example_patent.py) and diffed the resulting docx XML against a plain python-docx baseline: baseline Heading 1 carries Word's default blue theme color (w:themeColor=\"accent1\", RGB 365F91) and literal 'x_i'/'y_i' underscore text, while the skill's output has 18 real m:oMath subscript elements and zero themeColor references — exactly the two failure modes the skill's own pitfalls table calls out, both concretely fixed.", "testedDate": "2026-07-16", "repo": "https://github.com/StellaQIU27/claude-skill-patent-writing-chinese", "install": "git clone https://github.com/StellaQIU27/claude-skill-patent-writing-chinese\nmkdir -p ~/.claude/skills\ncp -r claude-skill-patent-writing-chinese ~/.claude/skills/patent-writing-chinese", "tags": [ "docx", "patent", "chinese", "python-docx", "ommls" ] }, { "slug": "yt-digest", "name": "YT Digest", "url": "https://skillproof.dev/skills/yt-digest", "category": "productivity", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Turns a YouTube URL into a timestamped, screenshotted markdown research note via real yt-dlp + ffmpeg.", "testNotes": "Ran the entire real pipeline (yt-dlp fetch, VTT parse, ffmpeg frame grab, markdown assembly, cleanup) end to end on a live YouTube video, no shortcuts: correct metadata, deduplicated captions, real extracted screenshots, ~21MB source video actually deleted afterward leaving only a 1.3MB vault note+images — it does exactly what it claims, with zero friction.", "testedDate": "2026-07-16", "repo": "https://github.com/tga-cheetung/yt-digest", "install": "git clone https://github.com/tga-cheetung/yt-digest.git\nmkdir -p ~/.claude/skills/yt-digest\ncp -r yt-digest/SKILL.md yt-digest/scripts ~/.claude/skills/yt-digest/", "tags": [ "youtube", "transcription", "research", "obsidian", "yt-dlp" ] }, { "slug": "html-explainer", "name": "HTML Explainer", "url": "https://skillproof.dev/skills/html-explainer", "category": "design", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Builds polished, source-grounded, playable single-page or slide-deck web explainers with a real quality loop.", "testNotes": "Real A/B, rendered in-browser: the skill build produced a genuinely playable Doppler-shift canvas demo with color-coded compressed/stretched wavefronts, editorial serif/sans typography, and a real assemble.mjs-injected review overlay, against a baseline that was exactly the emoji-hero/gradient-text/bento-card template the skill's own anti-slop checklist warns about.", "testedDate": "2026-07-16", "repo": "https://github.com/ds-vibe/html-explainer", "install": "git clone https://github.com/ds-vibe/html-explainer\nmkdir -p ~/.claude/skills\ncp -r html-explainer ~/.claude/skills/html-explainer", "tags": [ "html", "interactive-demos", "educational-content", "design-system", "playwright" ] }, { "slug": "compone", "name": "Compone", "url": "https://skillproof.dev/skills/compone", "category": "coding", "verdict": "pass", "score": 10, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 5 }, "tagline": "Python component framework for type-safe HTML/XML/RSS generation.", "testNotes": "Ran a real A/B against the actual pip-installed compone 0.4.1: my unguided baseline guessed a dominate-style API and crashed immediately (TypeError: 'module' object is not callable), while the skill's documented @Component/bracket-notation pattern ran correctly on the first try and produced exact expected HTML.", "testedDate": "2026-07-16", "repo": "https://github.com/kissgyorgy/coding-agents", "install": "git clone https://github.com/kissgyorgy/coding-agents\nmkdir -p ~/.claude/skills\ncp -r coding-agents/skills/compone ~/.claude/skills/compone", "tags": [ "python", "html-generation", "components", "compone", "markup" ] }, { "slug": "docx", "name": "DOCX — Word Documents", "url": "https://skillproof.dev/skills/docx", "category": "documents", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Create and edit real Word documents — headings, tables, tracked changes.", "testNotes": "Produced a 14-page report with a table of contents, numbered headings, styled tables and a letterhead. Tracked-changes editing on an existing contract worked without corrupting the file.", "testedDate": "2026-06-24", "repo": "https://github.com/anthropics/skills", "install": "git clone https://github.com/anthropics/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r skills/docx ~/.claude/skills/docx", "tags": [ "word", "docx", "reports", "contracts" ] }, { "slug": "meeting-notes-formatter", "name": "Meeting Notes to Actions", "url": "https://skillproof.dev/skills/meeting-notes-formatter", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Turns meeting transcripts into quote-cited action items - no guessed owners or due dates.", "testNotes": "Fed it a transcript with a corrected misassignment, a joke commitment ('I'll take care of world peace'), and a vague 'someone should update the roadmap, soon': it correctly left the roadmap item Unassigned with a null due date and routed the joke to skipped.md, while a naive read attributed the roadmap item to a random speaker and logged the joke as a real task. (Listed here under our older name; the skill's own name is \"Meeting Notes to Actions\".)", "testedDate": "2026-07-14", "repo": "https://github.com/Justin2259/claude-skill-meeting-notes-to-actions", "install": "git clone https://github.com/Justin2259/claude-skill-meeting-notes-to-actions\ncd claude-skill-meeting-notes-to-actions\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/meeting-notes-formatter", "tags": [ "meetings", "transcripts", "action-items", "productivity", "summarization" ] }, { "slug": "skill-creator", "name": "Skill Creator", "url": "https://skillproof.dev/skills/skill-creator", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "The skill that builds skills — scaffold, test and optimize your own.", "testNotes": "Scaffolded a working internal skill (frontmatter, triggers, reference files) in one session; the description-optimization step measurably improved triggering.", "testedDate": "2026-06-26", "repo": "https://github.com/anthropics/skills", "install": "git clone https://github.com/anthropics/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r skills/skill-creator ~/.claude/skills/", "tags": [ "skills", "authoring", "meta" ] }, { "slug": "test-driven-development", "name": "Test-Driven Development", "url": "https://skillproof.dev/skills/test-driven-development", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Forces Claude into strict red-green-refactor discipline.", "testNotes": "Enforced red-green-refactor across a 3-feature session; Claude wrote failing tests first every time and refused to skip the cycle. Noticeably fewer regressions.", "testedDate": "2026-06-25", "repo": "https://github.com/obra/superpowers", "install": "/plugin marketplace add obra/superpowers-marketplace\n/plugin install superpowers", "tags": [ "tdd", "testing", "discipline" ] }, { "slug": "systematic-debugging", "name": "Systematic Debugging", "url": "https://skillproof.dev/skills/systematic-debugging", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Stops guess-and-check fixes; enforces hypothesis-driven debugging.", "testNotes": "Root-caused a race condition Claude had previously \"fixed\" three times with guesses. The hypothesis-test-verify loop stopped the guessing.", "testedDate": "2026-06-25", "repo": "https://github.com/obra/superpowers", "install": "/plugin marketplace add obra/superpowers-marketplace\n/plugin install superpowers", "tags": [ "debugging", "discipline" ] }, { "slug": "code-review-checklist", "name": "Code Review Checklist", "url": "https://skillproof.dev/skills/code-review-checklist", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Structured PR review checklist: security, performance, correctness, severity-labeled.", "testNotes": "Ran it against a two-endpoint Express snippet: a naive pass caught only the obvious SQL-injection string interpolation, but the checklist surfaced four Critical findings the naive pass missed entirely — an IDOR on the orders route, an unauthenticated password-reset endpoint, plaintext password storage, and no rate limiting.", "testedDate": "2026-07-14", "repo": "https://github.com/LeoYeAI/openclaw-master-skills", "install": "git clone https://github.com/LeoYeAI/openclaw-master-skills\ncd openclaw-master-skills\nmkdir -p ~/.claude/skills\ncp -r skills/code-review ~/.claude/skills/code-review-checklist", "tags": [ "code-review", "pull-requests", "security", "checklist", "engineering-process" ] }, { "slug": "frontend-design", "name": "Frontend Design", "url": "https://skillproof.dev/skills/frontend-design", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 10, "docs": 4 }, "tagline": "Makes Claude's UI output look designed instead of generated.", "testNotes": "Same landing-page brief run with and without the skill: the with-skill version had a real type scale, intentional palette and none of the neon-gradient tells. Night and day.", "testedDate": "2026-06-26", "repo": "https://github.com/anthropics/skills", "install": "git clone https://github.com/anthropics/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r skills/frontend-design ~/.claude/skills/", "tags": [ "design", "css", "ui" ] }, { "slug": "ui-audit", "name": "WCAG 2.1 AA Web UI Audit", "url": "https://skillproof.dev/skills/ui-audit", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Deterministic WCAG 2.1 AA audit -- keyboard, contrast, zoom/reflow, forms/ARIA, status messages -- ending in a severity-rated remediation backlog.", "testNotes": "Ran the bundled `run_axe_playwright.js` for real with no playwright/axe installed: it printed install guidance and exited 0 instead of crashing, exactly as documented. On a 3-line login-form snippet it produced a WCAG-SC-cited, severity-ranked findings table (password field missing a programmatic label, low-contrast submit button) instead of the loose bullet list a baseline pass gives. (Listed here under our older name; the skill's own name is \"WCAG 2.1 AA Web UI Audit\".)", "testedDate": "2026-07-14", "repo": "https://github.com/dvcrn/openclaw-skills-marketplace", "install": "git clone https://github.com/dvcrn/openclaw-skills-marketplace\ncd openclaw-skills-marketplace\nmkdir -p ~/.claude/skills\ncp -r plugins/nicolas-m-design--wcag-21-aa-web-ui-audit/skills/wcag-21-aa-web-ui-audit ~/.claude/skills/ui-audit", "tags": [ "accessibility", "wcag", "a11y", "audit", "ui" ] }, { "slug": "seo-audit", "name": "SEO Audit", "url": "https://skillproof.dev/skills/seo-audit", "category": "seo", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Structured technical/on-page/content SEO audit that explicitly flags a real trap: static fetch tools can't see JS-injected schema markup.", "testNotes": "Best concrete finding: the skill catches a real false-positive trap before you make it -- it explicitly warns that web_fetch/curl strip client-side-injected JSON-LD, so 'no schema found' from a raw fetch is often wrong; a baseline pass without this skill would have reported that false negative with confidence.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/seo-audit ~/.claude/skills/seo-audit", "tags": [ "seo", "technical-seo", "audit", "core-web-vitals", "on-page-seo" ] }, { "slug": "programmatic-seo", "name": "Programmatic SEO", "url": "https://skillproof.dev/skills/programmatic-seo", "category": "seo", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "12-playbook framework for templated pages at scale (locations, comparisons, integrations, glossary...) with an explicit thin-content guardrail checklist.", "testNotes": "A baseline pass on the mini-task ('one page per job title, swap the title in the H1') walks straight into the exact anti-pattern the skill names as mistake #1 in its own Common Mistakes list -- the skill instead forces proprietary-data differentiation and a thin-content pre-launch checklist before any pages ship.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/programmatic-seo ~/.claude/skills/programmatic-seo", "tags": [ "programmatic-seo", "pseo", "templated-pages", "scaled-content", "seo" ] }, { "slug": "copywriting", "name": "Copywriting", "url": "https://skillproof.dev/skills/copywriting", "category": "marketing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Conversion copywriting for marketing pages -- headline/CTA formulas, an explicit weak-vs-strong CTA blacklist, and structured output with rationale.", "testNotes": "A baseline pass on the mini-task produced a headline and CTA ('Manage Your Projects Better' / 'Get Started') that land exactly on the two anti-patterns the skill's own Weak CTAs / vague-language rules tell you to avoid -- the skill-guided version replaced both with an outcome-specific headline and a CTA naming what the user actually gets.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/copywriting ~/.claude/skills/copywriting", "tags": [ "copywriting", "landing-page-copy", "cta", "conversion-copywriting", "marketing" ] }, { "slug": "ab-testing", "name": "A/B Testing", "url": "https://skillproof.dev/skills/ab-testing", "category": "marketing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Designs statistically valid experiments and runs a full ICE-scored experimentation program.", "testNotes": "Follows its own peeking rule to the letter: given a fake 3-day result with 95% confidence, it refused to call a winner and cited regression-to-the-mean and day-of-week effects rather than declaring victory, something the baseline answer didn't even think to flag.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/ab-testing ~/.claude/skills/ab-testing", "tags": [ "experimentation", "hypothesis-testing", "statistics", "growth", "ice-scoring" ] }, { "slug": "content-strategy", "name": "Content Strategy", "url": "https://skillproof.dev/skills/content-strategy", "category": "marketing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Plans topic clusters, content pillars and a weighted-priority editorial calendar grounded in buyer stage.", "testNotes": "Replaced a flat blog-topic brainstorm with buyer-stage-mapped clusters and a weighted 40/30/20/10 scoring table for prioritization — the one weak spot is references/headless-cms.md, whose CMS integration links point at a repo-level tools/ folder that isn't included in a standalone copy, but that's a narrow optional path, not the core workflow.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/content-strategy ~/.claude/skills/content-strategy", "tags": [ "topic-clusters", "editorial-calendar", "content-pillars", "buyer-journey", "content-planning" ] }, { "slug": "account-research", "name": "Lead Research", "url": "https://skillproof.dev/skills/account-research", "category": "sales", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Account/contact research: trigger events, org-chart power mapping, ICP scoring -- honest about what it can't verify.", "testNotes": "Given a fictional Series B account and asked to build a brief, it did not invent a plausible VP-of-Sales name the way an unguided baseline would -- it flagged that it can't access LinkedIn Sales Nav or ZoomInfo and told the rep exactly what to search for instead, matching the 'don't fabricate data to fill gaps' rule the SKILL.md states. (Listed here under our older name; the skill's own name is \"Lead Research\".)", "testedDate": "2026-07-14", "repo": "https://github.com/TheCraigHewitt/sales-skills", "install": "git clone https://github.com/TheCraigHewitt/sales-skills\ncd sales-skills\nmkdir -p ~/.claude/skills\ncp -r skills/lead-research ~/.claude/skills/account-research", "tags": [ "lead-research", "account-research", "b2b-sales", "prospecting", "icp" ] }, { "slug": "call-prep", "name": "Discovery Call", "url": "https://skillproof.dev/skills/call-prep", "category": "sales", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Discovery/first-call prep: SPIN-derivative question bank, PACT qualification scoring, multi-stakeholder tactics.", "testNotes": "Its own description scopes it to first/discovery calls, and that held up in testing: a 'prep for a QBR with an existing customer' prompt correctly did not trigger it, confirming the card's own concern that this isn't a drop-in replacement for generic 'any sales call' prep. (Listed here under our older name; the skill's own name is \"Discovery Call\".)", "testedDate": "2026-07-14", "repo": "https://github.com/TheCraigHewitt/sales-skills", "install": "git clone https://github.com/TheCraigHewitt/sales-skills\ncd sales-skills\nmkdir -p ~/.claude/skills\ncp -r skills/discovery-call ~/.claude/skills/call-prep", "tags": [ "discovery-call", "call-prep", "spin-selling", "qualification", "b2b-sales" ] }, { "slug": "pipeline-review", "name": "Pipeline Review", "url": "https://skillproof.dev/skills/pipeline-review", "category": "sales", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Weighted deal scoring, zombie-deal detection, and forced disqualification for pipeline reviews -- works on pasted deal data, not a live CRM.", "testNotes": "Given two toy deals, it correctly flagged the 70-day-old, single-threaded, three-weeks-silent one as a zombie with a one-week resurrect-or-kill deadline, versus a baseline that just said 'follow up with them' -- the scoring framework did real diagnostic work instead of generic commentary.", "testedDate": "2026-07-14", "repo": "https://github.com/TheCraigHewitt/sales-skills", "install": "git clone https://github.com/TheCraigHewitt/sales-skills\ncd sales-skills\nmkdir -p ~/.claude/skills\ncp -r skills/pipeline-review ~/.claude/skills/pipeline-review", "tags": [ "pipeline-review", "deal-scoring", "sales-ops", "crm-hygiene", "b2b-sales" ] }, { "slug": "proposal-builder", "name": "Proposal & Pricing", "url": "https://skillproof.dev/skills/proposal-builder", "category": "sales", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Proposal/SOW structure, three-tier outcome-named pricing, and a Mutual Action Plan -- outputs text/markdown, not a formatted document.", "testNotes": "Produces a markdown/text proposal with three-tier, outcome-named pricing and a Mutual Action Plan timeline, but it does not generate a DOCX or call the docx skill itself -- so the old card's claim that it 'depends on the DOCX skill for output quality' was the site's own recommendation, not something built into this SKILL.md. (Listed here under our older name; the skill's own name is \"Proposal & Pricing\".)", "testedDate": "2026-07-14", "repo": "https://github.com/TheCraigHewitt/sales-skills", "install": "git clone https://github.com/TheCraigHewitt/sales-skills\ncd sales-skills\nmkdir -p ~/.claude/skills\ncp -r skills/proposal-pricing ~/.claude/skills/proposal-builder", "tags": [ "proposal-pricing", "sow", "pricing-strategy", "b2b-sales", "deal-close" ] }, { "slug": "objection-handler", "name": "Objection Handling", "url": "https://skillproof.dev/skills/objection-handler", "category": "sales", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "ACRC-framework objection playbook and live/email scripts for B2B reps — covers price, ghosting, and champion-relayed objections, not just pricing pushback.", "testNotes": "Tested it on a champion-relayed objection ('my legal team has concerns about data handling') instead of the SKILL's own worked example — it correctly combined the Team-Based Objection Handling section (equip the champion, ask for direct access) with the Technical/compliance script, something a plain unscripted reply skipped entirely. (Listed here under our older name; the skill's own name is \"Objection Handling\".)", "testedDate": "2026-07-14", "repo": "https://github.com/TheCraigHewitt/sales-skills", "install": "git clone https://github.com/TheCraigHewitt/sales-skills\ncd sales-skills\nmkdir -p ~/.claude/skills\ncp -r skills/objection-handling ~/.claude/skills/objection-handler", "tags": [ "sales", "objection-handling", "b2b", "playbook", "sales-enablement" ] }, { "slug": "metrics-review", "name": "Metrics Review", "url": "https://skillproof.dev/skills/metrics-review", "category": "data", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Turns pasted or queried metrics into a structured weekly/monthly/quarterly review — scorecard, trend analysis, bright spots vs. concerns, recommended actions. Does not build the dashboard itself.", "testNotes": "Verdict forced to setup for the same ../../CONNECTORS.md external-file dependency as its sibling dashboard-builder skill. On a 3-metric weekly snapshot (WAU down, retention below target, conversion down, no known incidents) it explicitly flagged the missing-explanation caveat in a dedicated Context and Caveats section ('correlation is not causation... no known events') where a plain restatement of the numbers just said things looked soft.", "testedDate": "2026-07-14", "repo": "https://github.com/anthropics/knowledge-work-plugins", "install": "git clone https://github.com/anthropics/knowledge-work-plugins\ncd knowledge-work-plugins\nmkdir -p ~/.claude/skills\ncp -r product-management/skills/metrics-review ~/.claude/skills/metrics-review", "tags": [ "product-metrics", "reporting", "okrs", "retention", "analytics" ] }, { "slug": "weekly-review", "name": "Weekly Review", "url": "https://skillproof.dev/skills/weekly-review", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "GTD-style weekly review turning wins, blockers, and OKR notes into a structured writeup.", "testNotes": "Ran the bundled synthesizer script on a sample week: scattered wins, blockers, and OKR notes came back as a clean structured review with owner+mitigation fields per blocker and four self-coach prompts appended automatically — real script, real output, seconds to run.", "testedDate": "2026-07-14", "repo": "https://github.com/borghei/Claude-Skills", "install": "git clone https://github.com/borghei/Claude-Skills\ncd Claude-Skills\nmkdir -p ~/.claude/skills\ncp -r personal-productivity/weekly-review ~/.claude/skills/weekly-review", "tags": [ "gtd", "weekly-review", "okr", "productivity", "python-script" ] }, { "slug": "humanizer", "name": "Humanizer", "url": "https://skillproof.dev/skills/humanizer", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Strips 33 documented AI-writing tells from text via a draft-audit-final rewrite loop.", "testNotes": "Ran a sample AI-sounding product-launch paragraph through the draft→audit→final loop: the draft still had a stray em dash and a generic closer, and the mandatory second pass caught and fixed both — a step a single-pass edit skipped entirely.", "testedDate": "2026-07-14", "repo": "https://github.com/blader/humanizer", "install": "git clone https://github.com/blader/humanizer\ncd humanizer\nmkdir -p ~/.claude/skills\ncp -r . ~/.claude/skills/humanizer", "tags": [ "ai-detection", "editing", "prose", "writing-style", "humanize" ] }, { "slug": "newsletter-writer", "name": "Substack Ghostwriting & Content Optimization", "url": "https://skillproof.dev/skills/newsletter-writer", "category": "marketing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Substack-specific ghostwriting and growth workflow: voice guide, algorithm, monetization.", "testNotes": "It's a Substack-specific growth/ghostwriting workflow (algorithm mechanics, Notes strategy, monetization math), not a generic newsletter tool — and Phase 1 is a hard stop: it produced five subject-line variants and three distinct hooks for review before writing a single word of body copy, versus a baseline that just wrote one draft outright. (Listed here under our older name; the skill's own name is \"Substack Ghostwriting & Content Optimization\".)", "testedDate": "2026-07-14", "repo": "https://github.com/samber/cc-skills", "install": "git clone https://github.com/samber/cc-skills\ncd cc-skills\nmkdir -p ~/.claude/skills\ncp -r skills/substack-ghostwriting ~/.claude/skills/newsletter-writer", "tags": [ "substack", "newsletter", "ghostwriting", "content-marketing", "growth" ] }, { "slug": "blog-outliner", "name": "Blog Outline Generator", "url": "https://skillproof.dev/skills/blog-outliner", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "SERP-informed H2/H3 outline generator; skeleton only, not a full content brief.", "testNotes": "Live-tested on 'best project management software for remote teams': real SERP research surfaced the market's dominant 'best-for-X' comparison-table-plus-FAQ pattern (confirmed by fetching a competitor's actual H2/H3 structure), which the skill folded straight into the outline's content-gap section. A generic outline written without that research never would have known the pattern existed. (Listed here under our older name; the skill's own name is \"Blog Outline Generator\".)", "testedDate": "2026-07-14", "repo": "https://github.com/AgriciDaniel/claude-blog", "install": "git clone https://github.com/AgriciDaniel/claude-blog.git\ncd claude-blog\nmkdir -p ~/.claude/skills\ncp -r skills/blog-outline ~/.claude/skills/blog-outliner", "tags": [ "blogging", "outlines", "seo" ] }, { "slug": "schema-markup", "name": "Schema Markup", "url": "https://skillproof.dev/skills/schema-markup", "category": "seo", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "JSON-LD generator/validator for schema.org rich-results markup.", "testNotes": "Asked for schema on a SaaS pricing page: the skill combined Organization + SoftwareApplication + FAQPage under one @graph block using the correct required properties (name+offers for SoftwareApplication) straight from its own reference table. An unguided pass produced a single incomplete SoftwareApplication block that was missing offers entirely.", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills.git\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/schema ~/.claude/skills/schema-markup", "tags": [ "schema", "json-ld", "rich-results" ] }, { "slug": "geo-ai-visibility", "name": "AI SEO", "url": "https://skillproof.dev/skills/geo-ai-visibility", "category": "seo", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "GEO: AI-crawler access checks, extractable content structure, citation tracking.", "testNotes": "Ran a citability check for a SaaS 'X vs Y' comparison page: it named the exact bots to check for (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) and cited the '~33% of AI citations go to comparison articles' figure to justify a comparison-table structure — specific, checkable guidance instead of generic 'be more visible to AI' advice. (Listed here under our older name; the skill's own name is \"AI SEO\".)", "testedDate": "2026-07-14", "repo": "https://github.com/coreyhaines31/marketingskills", "install": "git clone https://github.com/coreyhaines31/marketingskills.git\ncd marketingskills\nmkdir -p ~/.claude/skills\ncp -r skills/ai-seo ~/.claude/skills/geo-ai-visibility", "tags": [ "geo", "ai-search", "llms-txt" ] }, { "slug": "semantic-seo-suite", "name": "Semantic SEO Suite", "url": "https://skillproof.dev/skills/semantic-seo-suite", "category": "seo", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Entity-based topical maps, briefs, and fabrication-guarded SEO drafts as Claude Code skills", "testNotes": "Fetched the repo tree (11 skills under skills/*/SKILL.md, 14 framework docs, 25 scripts, JSON schemas); tested the flagship skills/topical-map-builder/SKILL.md. Frontmatter parses with name+description; every referenced path (framework/topical-map-theory.md, eav-modeling.md, templates/topical-map.schema.json, scripts) exists in the tree. Grepped SerpApi/DataForSEO/GSC/autocomplete scripts — all clean, keys read from env only, no exfiltration, curl|sh, or hidden instructions. Measured T0 output: skill produced a tiered core/outer hub-and-spoke map with intent+provenance tags and cannibalization merges vs a flat baseline list — a clear, specific win satisfying the skill's own eval assertion. README is unusually honest (explicit \"not included yet\", refuses to publish unattributable outcome stats), so no docs overselling.", "testedDate": "2026-07-20", "repo": "https://github.com/siddiqss/semantic-seo-suite", "install": "/plugin marketplace add siddiqss/semantic-seo-suite then /plugin install semantic-seo-suite — or clone the repo and open in Claude Code. T0 needs nothing; optional `pip install jsonschema pyyaml` (+ sentence-transformers/trafilatura/google-api for T1, SERPAPI_KEY for T2).", "tags": [ "seo", "semantic-seo", "topical-map", "content-strategy", "aeo", "geo", "schema", "claude-skills" ] }, { "slug": "api-and-interface-design", "name": "API And Interface Design", "url": "https://skillproof.dev/skills/api-and-interface-design", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Checklist-driven REST/GraphQL and TypeScript interface design with error and naming rules", "testNotes": "Turned a verb-based API design (POST /createBookmark, inconsistent returns) into proper resource routes, one consistent error shape, cursor pagination and a branded ID type.", "testedDate": "2026-07-10", "repo": "https://github.com/addyosmani/agent-skills", "install": "git clone https://github.com/addyosmani/agent-skills\ncd agent-skills\nmkdir -p ~/.claude/skills\ncp -r skills/api-and-interface-design ~/.claude/skills/api-and-interface-design", "tags": [ "api-design", "rest", "graphql" ] }, { "slug": "screen-reader-testing", "name": "Screen Reader Testing", "url": "https://skillproof.dev/skills/screen-reader-testing", "category": "testing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Screen reader testing playbook for VoiceOver, NVDA, and JAWS with ARIA fixes.", "testNotes": "Given a modal with an icon-only close button and no dialog role, it produced the exact aria-label and role=dialog fixes plus concrete VoiceOver and NVDA test scripts a generic answer skipped entirely.", "testedDate": "2026-07-10", "repo": "https://github.com/wshobson/agents", "install": "git clone https://github.com/wshobson/agents\ncd agents\nmkdir -p ~/.claude/skills\ncp -r plugins/accessibility-compliance/skills/screen-reader-testing ~/.claude/skills/screen-reader-testing", "tags": [ "accessibility", "screen-reader", "aria" ] }, { "slug": "adaptyv", "name": "Adaptyv", "url": "https://skillproof.dev/skills/adaptyv", "category": "data", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Submit and track protein binding/stability assays via Adaptyv's Foundry API and Python SDK.", "testNotes": "An answer written from memory guessed the wrong base URL, sequence format and status string; following the docs matched the real API schema exactly, catching 4 concrete errors that would fail live.", "testedDate": "2026-07-10", "repo": "https://github.com/K-Dense-AI/scientific-agent-skills", "install": "git clone https://github.com/K-Dense-AI/scientific-agent-skills\ncd scientific-agent-skills\nmkdir -p ~/.claude/skills\ncp -r skills/adaptyv ~/.claude/skills/adaptyv", "tags": [ "bioinformatics", "api", "python-sdk" ] }, { "slug": "frontend-slides", "name": "Frontend Slides", "url": "https://skillproof.dev/skills/frontend-slides", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Builds zero-dependency, fixed-16:9 HTML decks or converts PPTX, avoiding AI-slop defaults.", "testNotes": "Built the same pitch deck twice: the plain version fell into purple-gradient-on-white with system fonts, the guided one used a real font pairing and a scaling stage that rendered with no overflow in a live browser.", "testedDate": "2026-07-10", "repo": "https://github.com/zarazhangrui/frontend-slides", "install": "git clone https://github.com/zarazhangrui/frontend-slides\ncd frontend-slides\nmkdir -p ~/.claude/skills\ncp -r plugins/frontend-slides/skills/frontend-slides ~/.claude/skills/frontend-slides", "tags": [ "html", "presentations", "pptx-conversion" ] }, { "slug": "shipping-artifacts", "name": "Shipping Artifacts", "url": "https://skillproof.dev/skills/shipping-artifacts", "category": "documents", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Defines the doc set that makes an AI-built app reviewable before it ships.", "testNotes": "For a Next.js + Supabase + Stripe app it produced a structured doc set with per-step deny cases and a permissions matrix instead of a flat README, and its own docs admit test docs need a separate command.", "testedDate": "2026-07-10", "repo": "https://github.com/phuryn/pm-skills", "install": "git clone https://github.com/phuryn/pm-skills\ncd pm-skills\nmkdir -p ~/.claude/skills\ncp -r pm-ai-shipping/skills/shipping-artifacts ~/.claude/skills/shipping-artifacts", "tags": [ "documentation", "code-review", "security" ] }, { "slug": "skill-security-auditor", "name": "Skill Security Auditor", "url": "https://skillproof.dev/skills/skill-security-auditor", "category": "testing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "OWASP-mapped code/secrets/config audit with working bundled scan scripts", "testNotes": "Run against a 13-line file with SQL injection, command injection and a hardcoded API key, its scripts caught all three plus a missing .gitignore issue that a manual review missed.", "testedDate": "2026-07-10", "repo": "https://github.com/eigent-ai/eigent", "install": "git clone https://github.com/eigent-ai/eigent\ncd eigent\nmkdir -p ~/.claude/skills\ncp -r resources/example-skills/skill-security-auditor ~/.claude/skills/skill-security-auditor", "tags": [ "security", "owasp", "secrets-detection" ] }, { "slug": "notebooklm", "name": "NotebookLM", "url": "https://skillproof.dev/skills/notebooklm", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Scriptable Google NotebookLM CLI: notebooks, sources, podcasts, quizzes, reports, slides.", "testNotes": "Recalling CLI commands from memory got the syntax wrong (new-notebook, add-source); following the docs produced the real commands, and they are unusually upfront about which generation types are rate-limit-prone.", "testedDate": "2026-07-10", "repo": "https://github.com/teng-lin/notebooklm-py", "install": "git clone https://github.com/teng-lin/notebooklm-py\nmkdir -p ~/.claude/skills/notebooklm && cp notebooklm-py/SKILL.md ~/.claude/skills/notebooklm/SKILL.md\npip install \"notebooklm-py[browser]\"", "tags": [ "notebooklm", "google", "research-automation" ] }, { "slug": "academic-paper-reviewer", "name": "Academic Paper Reviewer", "url": "https://skillproof.dev/skills/academic-paper-reviewer", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Simulates rigorous academic peer review across originality, methodology, results, and writing.", "testNotes": "Turned a freeform review paragraph into a structured 4-dimension writeup with Major/Minor tags and a priority table - markedly more actionable than the unguided version.", "testedDate": "2026-07-10", "repo": "https://github.com/zebbern/claude-code-guide", "install": "git clone https://github.com/zebbern/claude-code-guide\ncd claude-code-guide\nmkdir -p ~/.claude/skills\ncp -r skills/academic-paper-reviewer ~/.claude/skills/academic-paper-reviewer", "tags": [ "peer-review", "academic-writing", "research" ] }, { "slug": "frontend-forge-fi-operations", "name": "Frontend Forge FI Operations", "url": "https://skillproof.dev/skills/frontend-forge-fi-operations", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Operate KubeSphere FrontendIntegration resources and the frontend-forge extension via kubectl.", "testNotes": "On a Kubernetes-style task, general knowledge guessed a namespace flag that doesn't apply; the skill caught the real preflight extension check and used the exact status fields from its own docs.", "testedDate": "2026-07-10", "repo": "https://github.com/kubesphere/kubesphere", "install": "git clone https://github.com/kubesphere/kubesphere\ncd kubesphere\nmkdir -p ~/.claude/skills\ncp -r skills/frontend-forge-fi-operations ~/.claude/skills/frontend-forge-fi-operations", "tags": [ "kubesphere", "kubectl", "kubernetes" ] }, { "slug": "preline-theme-generator", "name": "Preline Theme Generator", "url": "https://skillproof.dev/skills/preline-theme-generator", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Generates full light/dark Preline CSS theme files from a hue or brand hex via a bundled local Node script.", "testNotes": "Ran the actual bundled script with zero setup and no network calls, producing a 682-line CSS file with a full OKLCH scale and dark-mode chart tokens, versus a 16-property hand-written sheet with none.", "testedDate": "2026-07-10", "repo": "https://github.com/htmlstreamofficial/preline", "install": "git clone https://github.com/htmlstreamofficial/preline\ncd preline\nmkdir -p ~/.claude/skills\ncp -r skills/theme-generator ~/.claude/skills/preline-theme-generator", "tags": [ "preline", "tailwind-css", "design-tokens" ] }, { "slug": "deepchat-data-import", "name": "DeepChat Data Import", "url": "https://skillproof.dev/skills/deepchat-data-import", "category": "data", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Build read-only importers/migration tools for DeepChat's encrypted SQLite chat data", "testNotes": "An unguided export plan hedged with 'likely under Application Support'; the skill produced exact per-OS paths, the real database pragma sequence, working table names and explicit redaction rules for API keys.", "testedDate": "2026-07-10", "repo": "https://github.com/ThinkInAIXYZ/deepchat", "install": "git clone https://github.com/ThinkInAIXYZ/deepchat\ncd deepchat\nmkdir -p ~/.claude/skills\ncp -r .agents/skills/deepchat-data-import ~/.claude/skills/deepchat-data-import", "tags": [ "sqlite", "electron", "data-migration" ] }, { "slug": "antfu", "name": "Antfu", "url": "https://skillproof.dev/skills/antfu", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Anthony Fu's opinionated ESLint, TypeScript, and monorepo conventions for JS/TS projects.", "testNotes": "For a new TS library setup it surfaced concrete opinions a generic answer wouldn't reach alone: a pure-ESM build config, a one-line ESLint setup, and an exact git-hooks config, versus generic husky boilerplate.", "testedDate": "2026-07-10", "repo": "https://github.com/antfu/skills", "install": "git clone https://github.com/antfu/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r skills/antfu ~/.claude/skills/antfu", "tags": [ "eslint", "typescript", "tooling-conventions" ] }, { "slug": "web-quality-audit", "name": "Web Quality Audit", "url": "https://skillproof.dev/skills/web-quality-audit", "category": "testing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Lighthouse-style audit across Performance, Accessibility, SEO, and Best Practices with severity tiers.", "testNotes": "Auditing a deliberately broken test page, it caught the charset and mixed-content issues a flat review missed, tagging every finding with file:line and a concrete fix instead of a bare list.", "testedDate": "2026-07-10", "repo": "https://github.com/addyosmani/web-quality-skills", "install": "git clone https://github.com/addyosmani/web-quality-skills\ncd web-quality-skills\nmkdir -p ~/.claude/skills\ncp -r skills/web-quality-audit ~/.claude/skills/web-quality-audit", "tags": [ "lighthouse", "web-audit", "accessibility" ] }, { "slug": "pr-writing-review", "name": "PR Writing Review", "url": "https://skillproof.dev/skills/pr-writing-review", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Extracts before/after writing fixes and style lessons from GitHub PR review comments.", "testNotes": "Against a real merged PR, its extraction script pulled 21 suggestion blocks and correctly tracked the target file through 3 renames; a raw API call returned the same comments flat and unstructured.", "testedDate": "2026-07-10", "repo": "https://github.com/evalstate/fast-agent", "install": "git clone https://github.com/evalstate/fast-agent\ncd fast-agent\nmkdir -p ~/.claude/skills\ncp -r examples/hf-toad-cards/skills/pr-writing-review ~/.claude/skills/pr-writing-review", "tags": [ "github", "pr-review", "writing-feedback" ] }, { "slug": "microsoft-azure-webjobs-extensions-authentication-events-dotnet", "name": "Microsoft Azure WebJobs Extensions Authentication Events (.NET)", "url": "https://skillproof.dev/skills/microsoft-azure-webjobs-extensions-authentication-events-dotnet", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Reference for building Entra ID custom authentication event Azure Functions in .NET", "testNotes": "For adding a claim to an ID token, recall from memory hedged on the exact type names; the docs gave the precise, compiling type names and trigger attribute verbatim, checked against the real published package version.", "testedDate": "2026-07-10", "repo": "https://github.com/microsoft/skills", "install": "git clone https://github.com/microsoft/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r .github/plugins/azure-sdk-dotnet/skills/microsoft-azure-webjobs-extensions-authentication-events-dotnet ~/.claude/skills/microsoft-azure-webjobs-extensions-authentication-events-dotnet", "tags": [ "azure-functions", "entra-id", "dotnet" ] }, { "slug": "loopy", "name": "Loopy", "url": "https://skillproof.dev/skills/loopy", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Designs, finds, audits and runs bounded AI-agent loops with evidence receipts, backed by a live Loop Library catalog.", "testNotes": "Ran the real 'Find a published loop' workflow against the live catalog.json (85 real entries, fetched from signals.forwardfuture.com) for a flaky-CI-test scenario and it surfaced loop #044 'The test stabilizer loop' with an almost verbatim useWhen match, exact URL and author — a baseline answer without the skill would have had to invent or hand-wave a recommendation instead of grounding it in real, live data.", "testedDate": "2026-07-10", "repo": "https://github.com/Forward-Future/loopy", "install": "git clone https://github.com/Forward-Future/loopy\ncd loopy\nmkdir -p ~/.claude/skills\ncp -r skills/loopy ~/.claude/skills/loopy", "tags": [ "ai-agent-workflows", "loop-library", "automation-design", "audit", "evidence-based" ] }, { "slug": "portaljs-add-chart", "name": "PortalJS Add Chart", "url": "https://skillproof.dev/skills/portaljs-add-chart", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Adds a gated recharts chart to a PortalJS dataset showcase page.", "testNotes": "Ran a real add-chart pass in a scaffolded PortalJS portal: npm install recharts@^2.15.0 succeeded and tsc --noEmit passed clean; the skill correctly gated the chart to one dataset's showcase while a from-scratch baseline rendered the chart on every dataset's page instead — exactly the bug the SKILL.md warns about.", "testedDate": "2026-07-10", "repo": "https://github.com/datopian/portaljs", "install": "git clone https://github.com/datopian/portaljs\ncd portaljs\nmkdir -p ~/.claude/skills\ncp -r skills/portaljs-add-chart ~/.claude/skills/portaljs-add-chart", "tags": [ "portaljs", "recharts", "nextjs", "dataviz", "chart" ] }, { "slug": "osint-methodology", "name": "OSINT Methodology", "url": "https://skillproof.dev/skills/osint-methodology", "category": "testing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Structured 5-stage external recon methodology with confidence levels, severity rubric, and OpSec rules.", "testNotes": "The companion `offensive-osint` skill it repeatedly points to for implementation detail actually exists in the same repo and loads clean; an authorized-recon mini-task run through it produced ROI-ranked pipeline stages, TENTATIVE/FIRM/CONFIRMED confidence tagging, and a WAF detection back-off ladder that a baseline answer never generated unprompted.", "testedDate": "2026-07-10", "repo": "https://github.com/elementalsouls/Claude-OSINT", "install": "git clone https://github.com/elementalsouls/Claude-OSINT\ncd Claude-OSINT\nmkdir -p ~/.claude/skills\ncp -r skills/osint-methodology ~/.claude/skills/osint-methodology", "tags": [ "osint", "red-team", "recon", "security", "bug-bounty" ] }, { "slug": "best-practices", "name": "Best Practices", "url": "https://skillproof.dev/skills/best-practices", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Modern web security, compatibility and code-quality patterns from Lighthouse best-practices audits.", "testNotes": "Fetched the repo tree (6 sibling skills), read skills/best-practices/SKILL.md fully plus the README. Frontmatter parses with name+description; the one relative link (../web-quality-audit/SKILL.md) resolves in the tree; manual install is a plain cp with no hardcoded paths; no scripts, no external CLI needed to use it. Content is current and accurate (2024 polyfill.io attack, Trusted Types Baseline, prototype-pollution guidance, hidden-source-map + sourcesContent stripping). Only caveat: the \"security audit\" trigger phrase is broad enough to collide with non-web/backend contexts.", "testedDate": "2026-07-20", "repo": "https://github.com/addyosmani/web-quality-skills", "install": "cp -r skills/* ~/.claude/skills/ (or: npx add-skill addyosmani/web-quality-skills / plugin install)", "tags": [ "web", "security", "lighthouse", "code-quality", "frontend", "csp", "best-practices" ] }, { "slug": "golang-concurrency", "name": "Golang Concurrency", "url": "https://skillproof.dev/skills/golang-concurrency", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Turns hand-rolled goroutine code into leak-free errgroup pipelines with bounded, cancellable workers.", "testNotes": "Verified by execution, not just reading.", "testedDate": "2026-07-20", "repo": "https://github.com/samber/cc-skills-golang", "install": "Claude Code plugin. Add the samber/cc-skills-golang marketplace and install the plugin, or copy skills/golang-concurrency/ (SKILL.md + references/) into .claude/skills/. Requires the `go` toolchain on PATH for the workflows it guides (go test -race, vet); the skill guidance itself is self-contained.", "tags": [ "golang", "concurrency", "goroutines", "errgroup", "channels", "code-review" ] }, { "slug": "cold-start-problem", "name": "Cold Start Problem", "url": "https://skillproof.dev/skills/cold-start-problem", "category": "marketing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Andrew Chen's network-effects playbook as a scored launch framework for marketplaces and social apps", "testNotes": "Fetched SKILL.md and all 5 reference files raw from wondelai/skills@main; frontmatter parses with name+description, and every referenced references/*.md file exists in the git tree and is substantive. Grep of skill + all references found no curl|sh, base64, exfil, secrets, or hidden instructions — only 3 disclosed Amazon affiliate tags (tag=wondelai00-20) on Further Reading book links. Trigger phrasings — ACTIVATE: \"how do I get the first users for my two-sided marketplace?\", \"my social app has a chicken-and-egg problem, nobody's posting\", \"we're launching a marketplace with no buyers or sellers yet\"; NOT: \"write me a cold email sequence for B2B sales\" (cold ≠ cold-start), \"how do I improve my landing page conversion rate?\" — 5/5 correct, description is well-scoped and cross-references sibling skills to avoid over-triggering.", "testedDate": "2026-07-20", "repo": "https://github.com/wondelai/skills", "install": "Copy the cold-start-problem/ directory into .claude/skills/ (or .agents/skills/, etc. — the repo ships all five agent variants). Self-contained: SKILL.md plus 5 reference markdown files, no external CLI, MCP, or runtime required.", "tags": [ "network-effects", "marketplace", "growth", "product-launch", "startup-strategy", "atomic-network" ] }, { "slug": "pixel2motion", "name": "Pixel2motion", "url": "https://skillproof.dev/skills/pixel2motion", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Turns a raster logo into a minimal SVG, then a choreographed 12-principles motion reveal.", "testNotes": "Fetched the tree via gh (raw GitHub was rate-limited): one root SKILL.md (28KB), plus 5 references/*.md and 11 scripts/*.py — all present, and render_overlay.py's only subprocess call is headless Chrome, no exfiltration/hidden instructions. Measured OUTPUT for real: headless-Chrome screenshots show the skill version cascades parts with a draw-on ring + overshoot pop and lands cleanly on the final frame (3 literal cubic-beziers, 0 var() inside @keyframes, pathLength draw-on, reduced-motion block), while the baseline fades/scales everything in lockstep and linear. Trigger phrasings — SHOULD: 'animate my company logo into a splash screen reveal', 'turn this PNG logo into an animated SVG with a draw-on', 'create a loading animation for our brand mark'; SHOULD NOT: 'vectorize this landscape photo into an SVG', 'build a full CSS page-transition library for my site' — all 5 judged correctly.", "testedDate": "2026-07-20", "repo": "https://github.com/nolangz/pixel2motion", "install": "Copy the repo into ~/.claude/skills/pixel2motion (or a project .claude/skills dir). Core motion authoring works immediately; to run the bundled QA scripts install Pillow+numpy (venv), headless Chrome (CHROME_BIN), and Playwright.", "tags": [ "logo-animation", "svg", "css-keyframes", "motion-design", "disney-12-principles", "vectorization", "brand-reveal" ] }, { "slug": "convex-backend", "name": "Convex Backend", "url": "https://skillproof.dev/skills/convex-backend", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Convex backend rules: new function syntax, validators, indexed queries over filters.", "testNotes": "Fetched skills/convex-backend/SKILL.md (119 lines) and its referenced AGENTS.md (884 lines) raw from main; both exist, frontmatter parses with name+description, and a grep for curl/base64/eval/tokens/referral found nothing. Trigger phrasings tested — YES: \"write a Convex mutation to insert a user doc\", \"add an index to my Convex schema and query messages by channel\", \"set up a cron in my Convex backend to purge old records\"; NO: \"set up a Prisma schema for Postgres\", \"write a React component to render a message list\" — all 5 judged correctly. On the mini-task the baseline used .filter() (full table scan) and omitted the returns validator; the skill produced a schema .index() + .withIndex() indexed lookup with returns: v.null()/v.array(...), a genuine correctness/perf win.", "testedDate": "2026-07-20", "repo": "https://github.com/CloudAI-X/claude-workflow-v2", "install": "Copy the skills/convex-backend/ folder (SKILL.md + AGENTS.md) into your project's skills directory. No external CLI, MCP, or binary required; it is pure guidance that activates on Convex backend tasks.", "tags": [ "convex", "backend", "typescript", "serverless", "database", "coding" ] }, { "slug": "arxiv-preflight", "name": "Arxiv Preflight", "url": "https://skillproof.dev/skills/arxiv-preflight", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Grounded pre-submission check that catches fabricated references and LLM residue before arXiv upload.", "testNotes": "Fetched SKILL.md + all 4 scripts + 4 reference docs via sparse checkout and ran the real pipeline on the repo's own `hallucinated_ref` fixture. verify_references.py hit live OpenAlex/Crossref and correctly flagged both fabricated cites (liang2023quantum, nguyen2024recursive) as BLOCKER with counter-evidence (they resolve to unrelated real papers), matched the 3 genuine refs, and caught two year-metadata errors; generate_preflight_report.py produced a HOLD report with file:line locations. Security: no curl|sh, no eval/base64; network limited to Crossref/arXiv/OpenAlex/Semantic Scholar via urllib; one subprocess call is `pdftotext` only. Minor rough edges: a SyntaxWarning and datetime.utcnow deprecation warnings, and the report script uses `--refs` (its own --help is correct).", "testedDate": "2026-07-20", "repo": "https://github.com/aipoch/medical-research-skills", "install": "Copy the `arxiv-preflight/` directory into your skills folder (e.g. `~/.claude/skills/`). Pure Python 3.9+ stdlib, no pip installs; `pdftotext` (poppler) only needed for the optional PDF input path. Reference verification needs outbound network to the scholarly APIs.", "tags": [ "academic-writing", "arxiv", "reference-verification", "research-integrity", "latex", "python" ] }, { "slug": "lore", "name": "Lore", "url": "https://skillproof.dev/skills/lore", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Mines your SpecStory coding histories into corroborated, outcome-labeled skill candidates you forge into real skills.", "testNotes": "Cloned the repo and actually ran the engine (pure Node stdlib, zero npm deps): index parsed 6 multi-agent sessions across 2 nested projects, and report surfaced the corroborated pattern \"build:run × go build ▸ go test ▸ golangci-lint run\" (3 sessions, 2✓/1✗ outcomes labeled from the user's next reply, exact commands from executed shell blocks, path:line refs, git author attribution) plus a verbatim pass-through summary — a clear, specific win over baseline transcript-guessing. Security scan clean: only local `git` via execFileSync for authorship, no network imports, no curl|sh, and the plan-validation hook correctly denies non-engine forge plans. All referenced scripts (mine-skills.mjs, hooks/validate-plan.mjs, workflows, lib/*) exist and `forged check` idempotency works. TRIGGER — SHOULD fire: \"what could I make into a skill from my sessions?\", \"mine my lore and forge skills from my history\", \"turn my last 30 days of coding into reusable skills\"; SHOULD NOT: \"write me a skill for deploying to Vercel\" (author-from-scratch, no mining), \"summarize what I did in yesterday's session\" (session summary, not skill extraction) — 5/5 correct. Only friction: requires the separate SpecStory recorder to have captured sessions, but the skill+engine themselves are fully self-contained and measured working. Exceptionally honest docs (names its own dated failure modes and hook-enforces the fix).", "testedDate": "2026-07-20", "repo": "https://github.com/specstoryai/getspecstory", "install": "Claude Code plugin: /plugin marketplace add specstoryai/getspecstory, then /plugin install lore. Or copy the lore/ dir into your skills path. Needs Node >=22.5 (built-in SQLite); no external npm deps. Requires SpecStory to have recorded sessions to mine.", "tags": [ "skill-authoring", "mining", "specstory", "workflow-extraction", "claude-code-plugin", "cross-agent" ] }, { "slug": "workthreads", "name": "Workthreads", "url": "https://skillproof.dev/skills/workthreads", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Deterministic weekly rollup of team work threads from .specstory coding histories.", "testNotes": "Fetched workthreads/SKILL.md + all 6 lib scripts from the dev branch and ran the engine against the repo's own fixtures: `index` ingested 6 sessions across 2 projects, `threads` grouped them per-repo into New/Open/Recently-closed with exact path:line evidence, and correctly flagged the PAYMENT_RETRY thread as reverted:true from a git rollback. Self-contained (node:sqlite built-in, zero deps), git called only via execFileSync with fixed array args, and it ships deterministic secret-redaction (AWS/GitHub/Slack/OpenAI/Anthropic/JWT/bearer patterns) before any output — no network, no curl|sh, no exfiltration found; 11/11 bundled tests pass on Node 24.7.", "testedDate": "2026-07-20", "repo": "https://github.com/specstoryai/getspecstory", "install": "Clone the repo, then `bash workthreads/install.sh` (copies the engine to ~/.agents/skills/workthreads and symlinks into ~/.claude/skills). Requires Node >=22.5 for the built-in node:sqlite. Zero npm dependencies.", "tags": [ "status-report", "team-rollup", "specstory", "deterministic", "node", "self-contained" ] }, { "slug": "forgecad", "name": "Forgecad", "url": "https://skillproof.dev/skills/forgecad", "category": "design", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Parametric CAD in .forge.js files, with connector-based placement and forgecad CLI checks", "testNotes": "In a temp HOME I ran `npm install -g forgecad` (0.13.0) and confirmed `forgecad run` works with no login, then built the same two-part desk cable clip twice. The baseline (no skill docs) crashed on `rotate(90,0,0)`, and after fixing from the error message it ran but rendered a full ring floating off the plate edge — scene bbox 60x45x18 mm, min [-30,-30,-3], i.e. 15 mm of overhang past a 30 mm-deep plate. The skill run, after loading the prescribed core/coordinate/positioning docs, used `pointAlong` + `connector.male/female` + `matchTo` and `verify.connectorDistance`, giving bbox 60x30x15 mm, min [-30,-15,0], `✓ Verifications: 1 pass, 0 fail`, and a render showing the arch correctly seated on the plate top between both bolt holes. All 5 spot-checked doc paths returned HTTP 200; no curl|sh, base64 blobs, or injection text anywhere in the skill tree; the only real friction is the mandatory global npm CLI and the unsubstituted {{SKILL_DIR}} placeholder when installing by git clone.", "testedDate": "2026-07-21", "repo": "https://github.com/KoStard/forgecad-public-kit", "install": "rm -rf /tmp/forgecad-src ~/.claude/skills/forgecad\ngit clone --depth 1 https://github.com/KoStard/forgecad-public-kit.git /tmp/forgecad-src\nmkdir -p ~/.claude/skills\ncp -R /tmp/forgecad-src/skills/forgecad ~/.claude/skills/forgecad\nsed -i '' \"s|{{SKILL_DIR}}|$HOME/.claude/skills/forgecad|g\" ~/.claude/skills/forgecad/SKILL.md 2>/dev/null || sed -i \"s|{{SKILL_DIR}}|$HOME/.claude/skills/forgecad|g\" ~/.claude/skills/forgecad/SKILL.md\n# The sed line is required: SKILL.md ships with a literal {{SKILL_DIR}} placeholder in every doc path.\n# Required external tool (the skill is inert without it): npm install -g forgecad\n# Verified working offline with no account: forgecad run , forgecad render 3d, forgecad inspect physical gaps.\n# Hosted features (forgecad login, project clone/push, studio project sync) need a ForgeCAD account.\n# Vendor alternative that also installs 10 companion workflow skills into ~/.agents/skills: forgecad skill install\n# (use `forgecad skill install --core-only` for just this skill; it substitutes {{SKILL_DIR}} itself).", "tags": [ "cad", "3d-modeling", "parametric", "javascript", "forgecad", "engineering" ] }, { "slug": "geo-review-html", "name": "Geo Review Html", "url": "https://skillproof.dev/skills/geo-review-html", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Renders self-contained HTML review forms from GEO briefs/drafts, exporting client input as JSON", "testNotes": "Cloned the referenced files (SKILL.md, scripts/render_html.py, templates/review.html, schema — all HTTP 200) and ran the script end-to-end in a temp dir across all three modes. brief mode split my test Markdown on REQUIRED-FILL blockquotes into slot textareas (data-1, quote-1) and appended the meta-only slot data-2; draft mode split on H2 into intro/sec-* sections; index mode scanned briefs/ and drafts/ meta.json into cards with correct slot/fill counts. Verified placeholders fully replaced, brand primary_color #0055ff injected as accent override, and the template's export builds exactly the schema's required object (brief_id, filled_at, filled_by, answers, comments, status). Baseline was a 486-byte hand-written static form; the skill output is an 18.9KB self-contained file with localStorage autosave (4 refs), JSON export/download (10), approve state machine (7), comment toggles (24), and slot-guidance details dropdowns (8) — none present in baseline. Clean: pure stdlib, HTML-escapes input, escapes closing-slash in the JSON blob; no security smells.", "testedDate": "2026-07-21", "repo": "https://github.com/ViryaZheng/recomby-geo", "install": "git clone --depth 1 https://github.com/ViryaZheng/recomby-geo.git /tmp/geo-review-html-src\nmkdir -p ~/.claude/skills\ncp -R /tmp/geo-review-html-src/plugins/recomby-geo/skills/geo-review-html ~/.claude/skills/geo-review-html\n# Pure Python 3 stdlib — no pip installs. templates/review.html and scripts/render_html.py travel inside the skill dir, so it runs standalone.\n# Run e.g.: python3 ~/.claude/skills/geo-review-html/scripts/render_html.py --mode brief --md brief.md --meta brief.meta.json --brand brand.json --out out.html\n# Modes: brief (fill-form) | draft (section comments + approve) | index (client folder entry page).\n# review_feedback.schema.json lives at plugins/recomby-geo/schemas/ in the repo (documentation only; the script does not load it).", "tags": [ "geo", "seo", "html", "client-review", "content-brief", "python", "json-export" ] }, { "slug": "step-parts", "name": "Step Parts", "url": "https://skillproof.dev/skills/step-parts", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Search the step.parts catalog and download checksum-verified STEP files for standard CAD parts", "testNotes": "Fetched SKILL.md and spot-checked all 3 referenced files (script, references/step-parts-api.md, agents/openai.yaml) — all HTTP 200. Ran the bundled download_step_part.py against the live api.step.parts: search returned the correct ISO 4762 M3x12 record, and --download saved a real 58826-byte valid ISO-10303-21 STEP file to disk with checksumVerified:true (sha256 confirmed with shasum). Baseline without the skill produces no valid STEP asset (would require a manual McMaster/TraceParts login or a fabricated file with no checksum); the skill produced a verified canonical file plus provenance URLs. Security scan of the 207-line script found no shell exec, base64, or secret handling — pure urllib.", "testedDate": "2026-07-31", "repo": "https://github.com/earthtojake/step.parts", "install": "git clone --depth 1 https://github.com/earthtojake/step.parts.git /tmp/step-parts-src\nmkdir -p ~/.claude/skills\ncp -R /tmp/step-parts-src/skills/step-parts ~/.claude/skills/step-parts\n# Requires: python3 (stdlib only — urllib, hashlib; no pip installs)\n# Live hosted API at https://api.step.parts (no key). Downloader: ~/.claude/skills/step-parts/scripts/download_step_part.py\n# Example: python3 ~/.claude/skills/step-parts/scripts/download_step_part.py \"M3 socket head 12\" --download --out-dir /tmp/step-parts", "tags": [ "cad", "step-file", "mechanical", "fasteners", "hardware", "catalog-search", "api", "python" ] }, { "slug": "alpaca-broker-account-onboarding", "name": "Alpaca Broker Account Onboarding", "url": "https://skillproof.dev/skills/alpaca-broker-account-onboarding", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Open and manage Alpaca brokerage accounts: KYC, agreements, documents, status lifecycle.", "testNotes": "Asked how to submit a create-account request and know when trading is allowed: the skill-guided answer produced the exact 11-state status lifecycle and the 'a 200 from POST does not mean tradable, gate on ACTIVE' lesson verbatim, while the unguided baseline only vaguely said 'wait for approval' with no field-level accuracy.", "testedDate": "2026-07-13", "repo": "https://github.com/alpacahq/alpaca-skills", "install": "git clone https://github.com/alpacahq/alpaca-skills\ncd alpaca-skills\nmkdir -p ~/.claude/skills\ncp -r skills/broker-api/account-onboarding ~/.claude/skills/alpaca-broker-account-onboarding", "tags": [ "alpaca", "broker-api", "kyc", "onboarding", "fintech-api" ] }, { "slug": "chapter-writing", "name": "Chapter Writing", "url": "https://skillproof.dev/skills/chapter-writing", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Outline-first chapter drafting that reads your whole story bible and updates continuity after writing.", "testNotes": "Ran a real side-by-side on a scratch story project: the no-skill baseline chapter opened with 'Rain hammered the tin roofs...' — the exact cliche the skill's own writing-guidelines.md tells you to avoid — while the skill-guided version opened on dialogue, planted a tracked setup/payoff in continuity/promises/_index.md, and left a full matching set of updated indices behind.", "testedDate": "2026-07-13", "repo": "https://github.com/danjdewhurst/story-skills", "install": "git clone https://github.com/danjdewhurst/story-skills\ncd story-skills\nmkdir -p ~/.claude/skills\ncp -r skills/chapter-writing ~/.claude/skills/chapter-writing", "tags": [ "fiction-writing", "story-continuity", "chapter-outline" ] }, { "slug": "critical-perspective", "name": "Critical Perspective", "url": "https://skillproof.dev/skills/critical-perspective", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Reframes one-sided conversations into open questions that surface hidden assumptions and gaps.", "testNotes": "Given the same 'go 100% paid ads, organic takes too long' pitch, the plain baseline gave a pros/cons list and a recommendation; running the skill's assumption-surfacing technique instead produced a single open question exposing the unexamined 'speed vs. compounding' framing and the bottom-of-funnel-only reach gap that 100%-paid leaves uncovered — a genuinely different, more exploratory response, exactly as the description promises.", "testedDate": "2026-07-13", "repo": "https://github.com/infranodus/skills", "install": "git clone https://github.com/infranodus/skills\ncd skills\nmkdir -p ~/.claude/skills\ncp -r skill-critical-perspective ~/.claude/skills/critical-perspective", "tags": [ "critical-thinking", "prompting", "devils-advocate", "discourse-analysis", "infranodus" ] }, { "slug": "langfuse", "name": "Langfuse", "url": "https://skillproof.dev/skills/langfuse", "category": "coding", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Playbooks that tell Claude which Langfuse MCP tool to call first when debugging LLM traces.", "testNotes": "Turned a vague 'why did this error happen' ask into an exact, correctly-ordered langfuse-mcp call chain (get_user_sessions -> get_session_details -> find_exceptions -> get_exception_details) with right parameter units (age in minutes) — but nothing runs until you separately install langfuse-mcp and paste in real Langfuse API keys.", "testedDate": "2026-07-13", "repo": "https://github.com/avivsinai/langfuse-mcp", "install": "git clone https://github.com/avivsinai/langfuse-mcp\ncd langfuse-mcp\nmkdir -p ~/.claude/skills\ncp -r skills/langfuse ~/.claude/skills/langfuse", "tags": [ "langfuse", "observability", "mcp-debugging" ] }, { "slug": "algorithmic-art", "name": "Algorithmic Art", "url": "https://skillproof.dev/skills/algorithmic-art", "category": "design", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Builds seeded p5.js generative art with sliders in one branded, self-contained interactive HTML file.", "testNotes": "templates/viewer.html ships a fully-wired Anthropic-branded seed navigator (prev/next/random/jump, parameter sliders, PNG download) that the skill forces every sketch through, so the output is consistently richer and more reproducible than an ad-hoc p5.js sketch written from scratch.", "testedDate": "2026-07-13", "repo": "https://github.com/ThinkInAIXYZ/deepchat", "install": "git clone https://github.com/ThinkInAIXYZ/deepchat\ncd deepchat\nmkdir -p ~/.claude/skills\ncp -r resources/skills/algorithmic-art ~/.claude/skills/algorithmic-art", "tags": [ "p5.js", "generative-art", "creative-coding", "seeded-randomness", "html-artifact" ] }, { "slug": "exam-drill", "name": "Exam Drill", "url": "https://skillproof.dev/skills/exam-drill", "category": "productivity", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Generates exam-tier twin/blind/chain drills from a course you've already ingested and analyzed.", "testNotes": "Its twin-generation rules force pattern-invariant variants with gated answer reveals, so a plain assistant asked the same thing would just print a twin problem plus the answer immediately; but the skill only produces anything once /ingest and /analyze (separate sibling skills in the same plugin) have already populated course-index/patterns.md and coverage.md for a specific course.", "testedDate": "2026-07-13", "repo": "https://github.com/OPTIMETA/PAIDEIA", "install": "git clone https://github.com/OPTIMETA/PAIDEIA\ncd PAIDEIA\nmkdir -p ~/.claude/skills\ncp -r plugins/paideia/skills/exam-drill ~/.claude/skills/exam-drill", "tags": [ "education", "study-drilling", "math-physics", "claude-plugin", "exam-prep" ] }, { "slug": "task-profile", "name": "Task Profile", "url": "https://skillproof.dev/skills/task-profile", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Mines your Claude Code/Cowork session history into a task profile, coaching panel, and skill proposals.", "testNotes": "Statically read the fetched inventory.py/host_platform.py against the SKILL.md's platform-detection claims and they match verbatim, down to the same 'AEAD-encrypted at rest, no honest analysis path -> degrade' caveat for Antigravity appearing both in the doc and in the code comments - an unusually well-aligned claims-vs-code check for a self-described skill.", "testedDate": "2026-07-13", "repo": "https://github.com/techwolf-ai/ai-first-toolkit", "install": "git clone https://github.com/techwolf-ai/ai-first-toolkit\ncd ai-first-toolkit\nmkdir -p ~/.claude/skills\ncp -r plugins/ai-adoption/skills/task-profile ~/.claude/skills/task-profile", "tags": [ "self-analysis", "productivity-coaching", "session-mining", "dashboard", "skill-proposals" ] }, { "slug": "golang-continuous-integration", "name": "Golang Continuous Integration", "url": "https://skillproof.dev/skills/golang-continuous-integration", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Production-grade GitHub Actions CI for Go: test, lint, SAST, releases, dependency bots.", "testNotes": "The bundled test.yml asset already ships go-version matrix, `-race -shuffle=on`, and a `go mod tidy && git diff --exit-code` guard — pulling it beats hand-writing a workflow from memory, and it matches known Go CI best practice almost line for line.", "testedDate": "2026-07-13", "repo": "https://github.com/samber/cc-skills-golang", "install": "git clone https://github.com/samber/cc-skills-golang\ncd cc-skills-golang\nmkdir -p ~/.claude/skills\ncp -r skills/golang-continuous-integration ~/.claude/skills/golang-continuous-integration", "tags": [ "golang", "github-actions", "ci-cd", "goreleaser", "security-scanning" ] }, { "slug": "commit-work", "name": "Commit Work", "url": "https://skillproof.dev/skills/commit-work", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Splits mixed diffs into logically-scoped, Conventional-Commits-formatted commits.", "testNotes": "On a diff mixing a real feature with unrelated reformatting, baseline output was one lumped `git commit -m \"update files\"`; following the skill produced `git add -p` staging split into a `feat(api):` commit and a separate `style(utils):` commit with proper bodies.", "testedDate": "2026-07-13", "repo": "https://github.com/softaworks/agent-toolkit", "install": "git clone https://github.com/softaworks/agent-toolkit\ncd agent-toolkit\nmkdir -p ~/.claude/skills\ncp -r dist/plugins/commit-work/skills/commit-work ~/.claude/skills/commit-work", "tags": [ "git", "conventional-commits", "version-control", "code-review" ] }, { "slug": "financial-analyzing", "name": "Financial Analyzing", "url": "https://skillproof.dev/skills/financial-analyzing", "category": "data", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Calculates margins, ROI, ROA/ROE from raw P&L numbers and drafts a benchmarked financial report.", "testNotes": "Actually ran scripts/calculate_ratios.py against a synthetic P&L (revenue $1.2M, net income $180K) -- it correctly returned 58.33% gross margin, 15% net margin, 7.2% ROA and 20% ROE in under a second, and the bundled profitability.md benchmarks flagged the ROE-vs-ROA spread as a leverage red flag a plain prompt-only baseline never surfaced.", "testedDate": "2026-07-13", "repo": "https://github.com/huangjia2019/claude-code-engineering", "install": "git clone https://github.com/huangjia2019/claude-code-engineering\ncd claude-code-engineering\nmkdir -p ~/.claude/skills\ncp -r 04-Skills/projects/03-financial-skill/.claude/skills/financial-analyzing ~/.claude/skills/financial-analyzing", "tags": [ "financial-analysis", "ratios", "roi", "reporting", "python-script" ] }, { "slug": "matlab-build-simbiology-model", "name": "Build SimBiology Models", "url": "https://skillproof.dev/skills/matlab-build-simbiology-model", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Dense SimBiology API/rules reference that prevents real ambiguous-species and deprecated-property errors.", "testNotes": "Asked both paths to wire a two-compartment PK model with a 'Drug' species in each compartment: the unguided baseline wrote unqualified species references that SimBiology rejects as ambiguous across compartments, while the skill's Rule 5 (always use Compartment.Species qualified names) produced reaction strings that are actually valid.", "testedDate": "2026-07-13", "repo": "https://github.com/matlab/matlab-agentic-toolkit", "install": "git clone https://github.com/matlab/matlab-agentic-toolkit\ncd matlab-agentic-toolkit\nmkdir -p ~/.claude/skills\ncp -r skills-catalog/computational-biology/matlab-build-simbiology-model ~/.claude/skills/matlab-build-simbiology-model", "tags": [ "simbiology", "matlab", "pharmacokinetics", "api-reference", "diagram-layout" ] }, { "slug": "night-market-change-control", "name": "Night Market Change Control", "url": "https://skillproof.dev/skills/night-market-change-control", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "PR/release governance playbook for this repo: change classification, exact gate order, and cited non-negotiables.", "testNotes": "The doc's own 'Provenance and maintenance' section ships re-verification grep commands and flags an unverified claim as unverified (an incident's exact restored-file count) instead of asserting it — checked the one verifiable count (10 constitution rules) live and it matched exactly.", "testedDate": "2026-07-13", "repo": "https://github.com/athola/claude-night-market", "install": "git clone https://github.com/athola/claude-night-market\ncd claude-night-market\nmkdir -p ~/.claude/skills\ncp -r .claude/skills/night-market-change-control ~/.claude/skills/night-market-change-control", "tags": [ "pr-process", "release-management", "pre-commit-gates", "governance", "constitution" ] }, { "slug": "aws-cloudformation-elasticache", "name": "AWS CloudFormation ElastiCache", "url": "https://skillproof.dev/skills/aws-cloudformation-elasticache", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "CloudFormation patterns for private, encrypted, Multi-AZ ElastiCache Redis/Memcached tiers.", "testNotes": "The bundled references/examples.md ships nine complete production templates (Multi-AZ Redis, cluster mode, multi-region replication, serverless) — genuinely deeper backing material than the SKILL.md alone lets on.", "testedDate": "2026-07-13", "repo": "https://github.com/giuseppe-trisciuoglio/developer-kit", "install": "git clone https://github.com/giuseppe-trisciuoglio/developer-kit\ncd developer-kit\nmkdir -p ~/.claude/skills\ncp -r plugins/developer-kit-aws/skills/aws-cloudformation/aws-cloudformation-elasticache ~/.claude/skills/aws-cloudformation-elasticache", "tags": [ "aws", "cloudformation", "elasticache", "redis", "infrastructure-as-code" ] }, { "slug": "dt-platform-costs", "name": "DT Platform Costs", "url": "https://skillproof.dev/skills/dt-platform-costs", "category": "data", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Turns Dynatrace DPS billing events into ranked cost-driver reports via DQL — never fabricates dollar figures.", "testNotes": "Hard-codes a 'never show cost_weight as a dollar figure' rule plus a verbatim pre-results disclaimer; run against its own worked example (62.3 TiB log ingest, 51.5M RUM sessions) it refuses the fabricated $ estimates a bare model would guess at.", "testedDate": "2026-07-13", "repo": "https://github.com/Dynatrace/dynatrace-for-ai", "install": "git clone https://github.com/Dynatrace/dynatrace-for-ai\ncd dynatrace-for-ai\nmkdir -p ~/.claude/skills\ncp -r skills/dt-platform-costs ~/.claude/skills/dt-platform-costs", "tags": [ "data", "observability", "dql", "billing", "dynatrace", "cost-analysis" ] }, { "slug": "kill-dev-process", "name": "Kill Dev Process", "url": "https://skillproof.dev/skills/kill-dev-process", "category": "efficiency", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Classifies and safely kills orphaned dev servers/browsers with a hard exclusion list for databases and the IDE.", "testNotes": "Ran the actual step-1 investigation commands on a live dev machine: they surfaced postgres on :5432 and Claude's own IDE Helper node processes mixed in with real dev servers - exactly the kind of thing a naive 'kill everything matching node|chrome' baseline would have blindly nuked, which the skill's safety-check step explicitly excludes.", "testedDate": "2026-07-13", "repo": "https://github.com/nokonoko1203/claude-code-settings", "install": "git clone https://github.com/nokonoko1203/claude-code-settings\ncd claude-code-settings\nmkdir -p ~/.claude/skills\ncp -r skills/kill-dev-process ~/.claude/skills/kill-dev-process", "tags": [ "dev-environment", "process-management", "port-cleanup", "macos", "devops" ] }, { "slug": "unoplat-code-confluence-cli", "name": "Unoplat Code Confluence CLI", "url": "https://skillproof.dev/skills/unoplat-code-confluence-cli", "category": "coding", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "CLI wrapper for the Unoplat Code Confluence app: repo onboarding and AGENTS.md generation with auto-PR.", "testNotes": "Fetched the real SKILL.md and its referenced CLI README, both resolve cleanly; the workflow correctly gates destructive `service destroy` and token/credential printing behind explicit user confirmation instead of just doing it.", "testedDate": "2026-07-13", "repo": "https://github.com/unoplat/unoplat-code-confluence", "install": "git clone https://github.com/unoplat/unoplat-code-confluence\ncd unoplat-code-confluence\nmkdir -p ~/.claude/skills\ncp -r skills/unoplat-code-confluence-cli ~/.claude/skills/unoplat-code-confluence-cli\n# also requires: uv tool install \"git+https://github.com/unoplat/unoplat-code-confluence.git#subdirectory=unoplat-code-confluence-cli\"", "tags": [ "cli", "docker", "agents-md", "repo-onboarding", "code-confluence" ] }, { "slug": "wiki-self-heal", "name": "Wiki Self-Heal", "url": "https://skillproof.dev/skills/wiki-self-heal", "category": "productivity", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Audits a Karpathy-style LLM wiki for orphans, phantom links and stale claims, on a review branch", "testNotes": "Built a throwaway 4-page test vault with a real orphan page and a broken [[typing]] wikilink, then ran the actual audit-only procedure end to end: it created a heal branch, ran the orphan/phantom-link scan scripts verbatim from the skill's own reference doc, and committed a correctly formatted audit report — no hand-waving required.", "testedDate": "2026-07-13", "repo": "https://github.com/NulightJens/ai-second-brain-skills", "install": "git clone https://github.com/NulightJens/ai-second-brain-skills\ncd ai-second-brain-skills\nmkdir -p ~/.claude/skills\ncp -r wiki-self-heal ~/.claude/skills/wiki-self-heal", "tags": [ "knowledge-base", "second-brain", "automation" ] }, { "slug": "article-writing", "name": "Article Writing", "url": "https://skillproof.dev/skills/article-writing", "category": "writing", "verdict": "pass", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "Bans 'rapidly evolving landscape' and 'game-changer', forces proof-first sections in long-form drafts", "testNotes": "Wrote the same Python-venv guide intro twice: the baseline opened with 'in today's rapidly evolving software landscape' and closed on 'let's dive in!' — exactly the two patterns the skill's own banned list names — while the skill version led with a runnable command and cut both.", "testedDate": "2026-07-13", "repo": "https://github.com/mturac/everything-openai-codex", "install": "git clone https://github.com/mturac/everything-openai-codex\ncd everything-openai-codex\nmkdir -p ~/.claude/skills\ncp -r .agents/skills/article-writing ~/.claude/skills/article-writing", "tags": [ "blogging", "content-writing", "voice" ] }, { "slug": "jadx", "name": "Jadx", "url": "https://skillproof.dev/skills/jadx", "category": "testing", "verdict": "setup", "score": 9.6, "scores": { "install": 5, "trigger": 5, "output": 9, "docs": 5 }, "tagline": "APK decompile + a systematic grep checklist for secrets, weak crypto, SQLi, and WebView holes.", "testNotes": "Planted 6 realistic vulns (hardcoded API key/password, MD5 hashing, string-concat SQL rawQuery, JS-enabled WebView with addJavascriptInterface, MODE_WORLD_READABLE prefs) in a fake decompiled tree: an ad hoc baseline grep caught only 2, the skill's documented Workflow-1 checklist caught all 6. Note: no files are referenced from the SKILL.md itself, but the jadx binary + a JRE are required and are not bundled -- neither was available in this sandbox, so decompilation itself (as opposed to the post-decompile grep workflow) went untested.", "testedDate": "2026-07-13", "repo": "https://github.com/BrownFineSecurity/iothackbot", "install": "git clone https://github.com/BrownFineSecurity/iothackbot\ncd iothackbot\nmkdir -p ~/.claude/skills\ncp -r skills/jadx ~/.claude/skills/jadx", "tags": [ "android", "apk", "decompiler", "security", "reverse-engineering" ] } ] }