Claude Skills
The open format is called Agent Skills and works in Claude Code, Codex, Cursor and other agents — most people know it as Claude Skills.
Every Agent Skill we could find on GitHub, deduplicated by content. 86 961 files from 2 125 authors, of which 68 678 are unique — the rest is the same skill repackaged into someone else's repository. For each one: what it weighs in tokens, whether it ships runnable scripts, and which MCP servers it needs.
>- Activate when designing or modifying CLI command surfaces, command help text, install/init/run flows, error wording, or first-run experience in the APM CLI -- even when the user does not say "UX" explicitly.
>- Use this skill to translate a classifier's in-place verdict into a precise, page-by-page work plan for the docs-sync panel. Activate after docs-impact-classifier returns verdict in_place; reads the candidate page list, fetches the actual page contents, narrows scope to specific sections within each page, and emits the per-page task brief the panel fans out against.
>- Use this skill when the docs-impact-classifier returns a structural verdict, signalling that the documentation TOC must change to accommodate the PR. Proposes TOC deltas (new pages, moves, merges) and emits new-page outline stubs that the doc-sync panel later fleshes out. Holds the 3-promise narrative (consume / produce / govern) and the persona ramps as hard constraints.
>- Use this skill whenever a pull request is opened, reopened, or synchronized in microsoft/apm to assess whether and how the documentation corpus must change to stay truthful with the proposed code change. Activate even when the PR title or body says nothing about docs -- the skill must run on every PR to detect silent drift between code and docs. Classifies impact as no-change, in-place edit (one to a few paragraphs), or structural change (new page or TOC reshape), then orchestrates a CDO + doc-writer + python-architect + editorial-owner + growth-hacker loop to produce a patch-ready advisory. Does NOT review code quality, security, or test coverage. Does NOT auto-merge or auto-push doc edits.
>- Use only as the composed drive-to-merge stage of an APM batch orchestrator (batch-bug-shepherd, apm-issue-autopilot) that has already selected ONE open pull request in microsoft/apm. Do NOT use for user-facing requests to triage issues, sweep a queue, or open PRs -- the parent orchestrator owns those. Spawn one shepherd-driver inline review, runs the apm-review-panel, folds (by default) every recommendation inside the PR's stated scope, pushes to the head branch or a superseding PR that preserves authorship via commit trailers, watches CI to green, and iterates under fixed caps until ready-to-merge, advisory-with-deferred, superseded, or blocked. Also provides the cross-PR conflict-resolution and mergeability-gate phase. This is NOT a standalone entrypoint.
>- Use this skill to write the PR description (PR body) for any pull request opened against microsoft/apm. Produces one self-sufficient (WHAT), Implementation (HOW), 1-3 validated mermaid diagrams, explicit trade-offs, validation evidence, and a How-to-test section -- with every WHY-claim backed by a verbatim quote from PROSE or Agent Skills. Activate when the user asks to "write a PR description", "draft a PR body", "open a PR", "fill in the PR template", or any equivalent.
>- Activate for changes to project positioning, release communication, community-facing artifacts, or breaking-change decisions in microsoft/apm. Triggers on README, MANIFESTO, PRD, CHANGELOG, release workflows, and issue templates.
>- Activate for OSS adoption work -- README conversion surfaces, quickstart, templates, release announcements, contributor funnel, story angles -- and any update to the maintained growth strategy at WIP/growth-strategy.md.
>- Activate when reviewing or modifying dependency resolution, lockfile schema, package downloaders, signature/integrity checks, file integration cleanup, or anything that could expose APM to dependency confusion, typosquatting, malicious packages, or token leakage.
> Activate when creating new modules, refactoring class hierarchies, introducing design patterns, or making changes spanning 3+ files in the APM CLI codebase.
>- Use this skill to run a four-panel adversarial advisory review on any pull request that touches the OpenAPM specification artifact (docs/src/content/docs/specs/openapm-*.md), its inline / sidecar JSON Schemas (docs/src/content/docs/specs/schemas/*.schema.json), or the conformance fixture seed (tests/fixtures/spec-conformance/**). The panel fans out to four spec-ecosystem reviewers (swagger-openapi-editor, oci-distribution-editor, pkgmgr-registry-contract-editor, w3c-tag-architect), each running in its own agent thread, and a spec-editor synthesizer that produces a fold-now / defer-v0.1.1 / defer-v0.2 / reject list plus a ship decision keyed off a 1..10 shocked_meter scale. The orchestrator is labels, no merge gating. The panel is advisory -- it surfaces findings, prioritizes folds, and renders a ship recommendation that the maintainer weighs.
> Activate when code touches token management, credential resolution, git auth flows, GITHUB_APM_PAT, ADO_APM_PAT, AuthResolver, HostInfo, AuthContext, or any remote host authentication -- even if 'auth' isn't mentioned explicitly.
> Use this skill when editing or creating CLI output, logging, warnings, error messages, progress indicators, or diagnostic summaries in the APM codebase. Activate whenever code touches console helpers (_rich_success, _rich_warning, _rich_error, _rich_info, _rich_echo), DiagnosticCollector, STATUS_SYMBOLS, CommandLogger, or any user-facing terminal output — even if the user doesn't mention "logging" or "UX" explicitly.
>- Activate when designing or modifying CLI command surfaces, command help text, install/init/run flows, error wording, or first-run experience in the APM CLI -- even when the user does not say "UX" explicitly.
>- Use this skill to run a holistic regrounding pass on the entire microsoft/apm documentation corpus against current source code, page-by-page, and emit surgical fixes for stale claims. Activate when the maintainer wants a WHOLE-CORPUS audit (not per-PR review) -- typical triggers include "audit the docs", "reground the corpus", "check every page against code", "pre-release docs sweep", "the docs have drifted everywhere", or "we just reshaped the TOC, find dead links". Wave-batched and S7-verified; scales to the full ~112-page corpus in ~10 minutes wall-time. This is a (triggered by a diff); this skill is per-corpus (triggered by a maintainer ask). They share agent personas, schemas, and the docs index, but their triggers MUST NOT collide. Does NOT auto-merge, does NOT push without maintainer review, and does NOT replace per-PR drift detection.
>- Use this skill whenever a pull request is opened, reopened, or synchronized in microsoft/apm to assess whether and how the documentation corpus must change to stay truthful with the proposed code change. Activate even when the PR title or body says nothing about docs -- the skill must run on every PR to detect silent drift between code and docs. Classifies impact as no-change, in-place edit (one to a few paragraphs), or structural change (new page or TOC reshape), then orchestrates a CDO + doc-writer + python-architect + editorial-owner + growth-hacker loop to produce a patch-ready advisory. Does NOT review code quality, security, or test coverage. Does NOT auto-merge or auto-push doc edits.
>- Activate for OSS adoption work -- README conversion surfaces, quickstart, templates, release announcements, contributor funnel, story angles -- and any update to the maintained growth strategy at WIP/growth-strategy.md.
> configuring, authoring, or troubleshooting AI-agent packages, dependencies, compilation, MCP servers, policy, or any `apm` CLI command.
>- Activate when reviewing or modifying dependency resolution, lockfile schema, package downloaders, signature/integrity checks, file integration cleanup, or anything that could expose APM to dependency confusion, typosquatting, malicious packages, or token leakage.
> Activate when creating new modules, refactoring class hierarchies, introducing design patterns, or making changes spanning 3+ files in the APM CLI codebase.
Copy and text skill for antislop. Use when writing or editing prose: headlines, tone, CTAs, and anti-AI-writing patterns. Load with the core.
UI and visual skill for antislop. Use when building or editing any interface: color, layout, components, motion. Load with the core.
Code comment hygiene for AI coding agents: remove generic AI-slop comments, keep the valuable ones, never touch the code.
Human and accessibility skill for antislop. Contrast, keyboard, focus, and states for real people. Includes the contrast checker.
Anti Slop: Rules for AI Coding Agents. The core filter. Load always to stop generic AI slop.
Mobile layout skill for antislop. Use for layouts that reflow across screen sizes, phone to desktop: grids, overflow, tap targets. Load with the core.
Create cinematic product videos from shot recipe cards, a validated template, and code/audio assets (Remotion + real page screenshots + 2.5D camera moves + beat-synced cuts + sound design). Use when the user asks to turn a frontend project or webpage into a product video, says "use video-shotcraft to make a video/promo", names the Ink Press template or asks to reproduce its effect, or wants a single shot card's motion. 用镜头配方卡 + 已验收模板 + 代码/音频资产制作电影感产品视频（Remotion + 真实页面截图 + 2.5D 运镜 + 节奏卡点 + 声音设计）。当用户要求"用 video-shotcraft 做视频/宣传片"、把前端项目/网页做成产品视频、点名 Ink Press 模板或要求复刻模板片效果，或要用镜头卡做单个动效镜头时使用。
Convert Markdown to WeChat Official Account HTML, inspect supported providers/themes/prompts, generate article images, create drafts, write with creator styles, prepare title suggestions, and remove AI writing traces.
Convert Markdown to WeChat Official Account HTML. Use this whenever the user wants WeChat article formatting, article preview, WeChat draft upload, image generation for articles, cover or infographic generation, image-post creation, writer-style drafting, title suggestions, AI trace removal, or current discovery of supported providers, themes, prompts, and layout modules.
Use when: the user asks to make or render charts with flint-chart, visualize tabular data, generate a ChartAssemblyInput, validate/render through MCP, or add Flint to a JS/TS project. Author the semantic spec, transform data before Flint when needed, install/import Flint only when executable code is needed, and reserve backend-specific style tweaks for after compiling from Flint.
Use when: the user asks to make or render charts with flint-chart, visualize tabular data, generate a ChartAssemblyInput, validate/render through MCP, or add Flint to a JS/TS project. Author the semantic spec, transform data before Flint when needed, install/import Flint only when executable code is needed, and reserve backend-specific style tweaks for after compiling from Flint.
Establish baseline-vs-candidate evaluation plans, delegate missing evaluations, compare validated results, and decide quantization feasibility. Use when the user asks to compare baseline vs quantized runs, explain an accuracy drop/regression, verify whether a quantized checkpoint is acceptable, or compare NEL/MLflow evaluation outputs. Do NOT use for generic single-model evaluation without comparison intent (use evaluation), live NEL status/debugging (use launching-evals), or generic MLflow browsing without a comparison goal (use accessing-mlflow).
Query and browse evaluation results stored in MLflow. Use when the user wants to look up runs by invocation ID, compare metrics across models, fetch artifacts (configs, logs, results), or set up the MLflow MCP server. ALWAYS triggers on mentions of MLflow, experiment results, run comparison, invocation IDs in the context of results, or MLflow MCP setup.
> Review EAGLE3 pipeline experiment logs from the launcher's experiments/ directory. Summarizes pass/fail status for all 4 tasks, diagnoses failures with root causes and fixes, and flags warnings. Use when the user asks to review job logs, check experiment results, or diagnose why a specific task failed.
> Add a new model to the EAGLE3 offline pipeline. Generates an hf_offline_eagle3.yaml launcher config for a new model checkpoint, choosing the right hidden state dump backend (TRT-LLM / HF / vLLM) and GPU configuration. Use when user wants to run EAGLE3 on a model that does not yet have a YAML in tools/launcher/examples/ or asks how to configure the pipeline for a new checkpoint.
Deterministic end-to-end driver for day-0 quantized-checkpoint releases — chains PTQ → evaluation → comparison with enforced gates between stages (the evaluation stage deploys the checkpoint itself), and returns a publish decision (ACCEPT / REGRESSION / ANOMALOUS / INFEASIBLE). Use when the user asks to "release a model at day-0", "quantize and validate model X is within N% of baseline and tell me if it's publishable", or "run the full day-0 workflow". Do NOT use for single-stage requests — quantizing only (use ptq), serving only (use deployment), evaluating only (use evaluation), or comparing two existing runs (use compare-results).
Serve a quantized or unquantized LLM checkpoint as an OpenAI-compatible API endpoint using vLLM, SGLang, or TRT-LLM. Use when user says "deploy model", "serve model", "start vLLM server", "launch SGLang", "TRT-LLM deploy", "AutoDeploy", "benchmark throughput", "serve checkpoint", or needs an inference endpoint from a HuggingFace or ModelOpt-quantized checkpoint. Do NOT use for quantizing models (use ptq) or evaluating accuracy (use evaluation).
>- Inspect Hugging Face decoder layers on meta tensors and plan or run per-rank BF16, FP8, and NVFP4 GEMM or fused-MoE microbenchmarks with the bundled scripts and a local FlashInfer checkout. Use when choosing a model, GPU, TP, EP, or M/token-concurrency sweep; deriving common fused QKV and gate/up shapes without loading checkpoint weights; or using FlashInfer benchmark utilities. Do not use for end-to-end server throughput or request latency.
Run commands inside a remote Docker container via the file-based command relay (tools/debugger). Use when the user says "run in Docker", "run on GPU", "debug remotely", "run test in container", "check nvidia-smi", "run pytest in Docker", or needs to execute any command inside a Docker container that shares the repo filesystem. Requires the user to have started server.sh inside the container first.
Monitor submitted jobs (PTQ, evaluation, deployment) on SLURM clusters. Use when the user asks "check job status", "is my job done", "monitor my evaluation", "what's the status of the PTQ", "check on job <slurm_job_id>", or after any skill submits a long-running job. Also triggers on "nel status", "squeue", or any request to check progress of a previously submitted job.
> Validate that an EAGLE3 pipeline run completed successfully end-to-end. Checks all 4 steps produced expected artifacts, verifies acceptance rate meets threshold (>= 2.1), and produces a summary report. Use when user wants to verify a pipeline run or check benchmark results.
Evaluates accuracy of quantized or unquantized LLMs using NeMo Evaluator Launcher (NEL). Triggers on "evaluate model", "benchmark accuracy", "run MMLU", "evaluate quantized model", "run nel". Handles deployment, config generation, and evaluation execution. Not for quantizing models (use ptq), deploying/serving models (use deployment), or comparing completed baseline-vs-quantized results (use compare-results).
> Triage a failed EAGLE3 pipeline run. Identifies which step failed (data synthesis, hidden state dump, training, or benchmark), diagnoses root cause from logs, and suggests fixes. Use when user reports an EAGLE3 pipeline failure or asks why a specific step failed. Also helps debug new model support issues.
Run, monitor, analyze, and debug LLM evaluations via nemo-evaluator-launcher. Covers running evaluations, checking status and live progress, debugging failed runs, exporting artifacts and logs, and analyzing results. ALWAYS triggers on mentions of running evaluations, checking progress, debugging failed evals, analyzing or analysing runs or results, run directories or artifact paths on clusters, Slurm job issues, invocation IDs, or inspecting logs (client logs, server logs, SSH to cluster, tail logs, grep logs). Do NOT use for creating or modifying evaluation configs.
>- Run explicitly requested ModelOpt Quantization-Aware Distillation (QAD) on Slurm through Megatron Bridge to recover a measured BF16-to-PTQ accuracy gap. Use only when the user explicitly asks for QAD, including its topology, data preparation, Slurm launch, resume, checkpoint export, or recovery decisions.
>- Use when the user asks to find, search for, or optimize the best quantization recipe for a model, including direct requests like "find the best quantization choose compute-vs-memory success metrics, select ModelOpt recipe baselines, design AutoQuant/manual recipe deltas, interpret sensitivity, and decide next candidates. Do NOT use for a single known PTQ recipe run (use ptq), serving (use deployment), creating/running evals (use evaluation or launching-evals), monitoring jobs (use monitor), MLflow browsing (use accessing-mlflow), or comparing completed baseline-vs-candidate scores only (use compare-results).
>- Use when the user asks to "quantize a model", "run PTQ", "post-training quantization", "NVFP4 quantization", "FP8 quantization", "INT8 quantization", "INT4 AWQ", "quantize LLM", "quantize MoE", "quantize VLM", or needs to produce a quantized HuggingFace checkpoint from a pretrained model using ModelOpt. Do NOT use for multi-candidate recipe exploration or optimization (use quant-recipe-search).
Cherry-pick merged PRs labeled for a release branch into that branch, then open a PR and apply the cherry-pick-done label. Use when asked to "cherry-pick PRs for release/X.Y.Z", "pick PRs to release branch", or "cherry-pick labeled PRs".
Use when styling UI for discord.com — references the extracted design system tokens instead of inventing colors, spacing, or typography.
Extract the full design language from any website URL. Produces 8 output files including AI-optimized markdown, visual HTML preview, Tailwind config, React theme, shadcn/ui theme, Figma variables, W3C design tokens, and CSS variables. Also runs WCAG accessibility scoring. Use when user says 'extract design', 'get design system', 'design language', 'design tokens', 'what colors/fonts does this site use', or '/extract-design'.
Use when styling UI for clerk.com — references the extracted design system tokens instead of inventing colors, spacing, or typography.
Use when styling UI for webflow.com — references the extracted design system tokens instead of inventing colors, spacing, or typography.
> Cross-model second opinion from Google Gemini — a different AI reviewing the (pass/fail gate for Google Ads campaigns, SEO metadata, or code), challenge (adversarial stress-test that tries to break your changes), and consult (open Q&A with Gemini on Google Ads strategy, SEO best practices, or implementation questions). Use when the user says "gemini review", "ask gemini", "gemini challenge", "second opinion from gemini", "consult gemini", "stress test with gemini", "what would gemini say", "cross-model review", useful for Google Ads changes, SEO metadata updates, campaign structure decisions, keyword strategies, and bid/budget changes — Gemini has native Google ecosystem knowledge that complements Claude's analysis.
Google Ads account audit and business context setup. Run this first — it gathers business information, analyzes account health, and saves context that all other ads skills reuse. Trigger on "audit my ads", "ads audit", "set up my ads", "onboard", "account overview", "how's my account", "ads health check", "what should I fix in my ads", or when the user is new to NotFair and hasn't run an audit before. Also trigger proactively when other ads skills detect that business-context.json is missing.
Generate and A/B test Google Ads copy. Use when asked to write ad copy, headlines, descriptions, create ad variants, test ad messaging, improve CTR, or generate RSA (Responsive Search Ad) components. Trigger on "ad copy", "write ads", "headlines", "descriptions", "RSA", "responsive search ad", "ad text", "ad creative", "improve CTR", "ad A/B test", "ad variants", "write me an ad", "ad variation experiment", or when the user wants to improve click-through rate on existing ads.
Score and diagnose Google Ads landing pages. Use when asked to audit a landing page, check landing page quality, diagnose high-CTR but low-conversion-rate ad groups, improve Quality Score's Landing Page Experience component, or compare an ad group's messaging against its landing page. Trigger on "landing page audit", "landing page score", "landing page quality", "why is my conversion rate low", "LPX", "landing page experience", "ad to page match", or when `/google-ads-audit` surfaces a high-CTR / low-CVR ad group.
Manage Google Ads — performance, keywords, bids, budgets, negatives, campaigns, ads, search terms, QS, location targeting, bulk operations, experiments, asset management, portfolio bidding, offline conversions. Use for any mention of Google Ads, CPA, ROAS, ad spend, or campaign settings.
Meta Ads (Facebook + Instagram) account audit and business context setup. Run this first — it gathers business information, analyzes account health, and saves context that all other Meta ads skills reuse. Trigger on "audit my Meta ads", "audit my Facebook ads", "Meta ads audit", "set up my Meta ads", "onboard Meta", "Meta account overview", "how's my Meta account", "Meta health check", "what should I fix in my Facebook ads", or when the user is new to NotFair Meta and hasn't run an audit before. Also trigger proactively when other Meta ads skills detect that meta business-context.json is missing.
Manage Meta Ads (Facebook + Instagram) — performance, ROAS, CPM, frequency, audience overlap, learning phase, creative fatigue, budgets, ad sets, campaigns, ads. Use for any mention of Meta Ads, Facebook Ads, Instagram Ads, ROAS, CPM, ad spend, or campaign settings on Meta.
> Upgrade the NotFair plugin to the latest version. Updates the marketplace repo, installs the new version to the plugin cache, and updates installed_plugins.json. Use when asked to "upgrade notfair", "update notfair", or "get latest version". Also handles inline upgrade prompts when a skill detects UPGRADE_AVAILABLE at startup.
Claude Skills — questions
Answers built from the skills we actually parsed.