GITHUB AGENT SKILLS / 技能收录
给你的 AI 助手
装上专业技能
我们从 GitHub 持续收录、审核并分类优质的 Agent Skills,按职业与使用场景整理,复制一行命令即可安装到 Claude Code、Cursor 等智能体。
按职业浏览 Agent Skills
为不同岗位挑选的高频技能
按使用场景浏览 Agent Skills
从你要完成的任务出发
按创作者浏览 Agent Skills
按 GitHub Stars 排序的优秀创作者
Microsoft
@microsoftopenclaw
@openclawAnthropic
@anthropicsOpenAI
@openaiThe Apache Software Foundation
@apacheGitHub
@githubAutoGPT
@Significant-GravitasGarry Tan
@garrytan全部 Skills 482556
FFmpeg Format Conversion
Convert media files between formats - video containers, audio formats, and codec transcoding
reproduce-bug
Reproduce a bug in an isolated, production-mirroring sandbox and capture a failing regression test as the objective baseline to fix against. Saves a report to docs/reproductions/. Triggers on: "reproduce this bug", "can't reproduce", "make it fail reliably", "repro".
FFmpeg Audio Processing
Extract, normalize, mix, and process audio tracks - audio manipulation and analysis
release-conductor
Gate the open release-please PR before merging: verify CI green, feature PRs landed, no release blockers, changelog reflects intent. One-word approval, then merge. Reports the resulting release.yml run URL. release-please owns the changelog/bump/tag; release.yml owns OIDC npm publish — this skill only gates the merge in between. Triggers on: "ship a release", "cut a release", "merge the release PR", "is the release ready", "verify the release".
nanogpt-training
Train GPT-2 scale models (~124M parameters) efficiently on a single GPU. Covers GPT-124M architecture, tokenized dataset loading (e.g., HuggingFace Hub shards), modern optimizers (Muon, AdamW), mixed precision training, and training loop implementation.
modal-gpu
Run Python code on cloud GPUs using Modal serverless platform. Use when you need A100/T4/A10G GPU access for training ML models. Covers Modal app setup, GPU selection, data downloading inside functions, and result handling.
quality-review
Check changed code against the resolved dotbabel quality policy, then review semantic risks that tools cannot judge. Triggers on: "quality review", "quality check", "review code quality".
mhc-algorithm
Implement mHC (Manifold-Constrained Hyper-Connections) for stabilizing deep network training. Use when implementing residual connection improvements with doubly stochastic matrices via Sinkhorn-Knopp algorithm. Based on DeepSeek's 2025 paper (arXiv:2512.24880).
pulumi-specialist
Deep-dive Pulumi stack review, component design, Automation API audit, and secrets management. Use for structured investigations of Pulumi stack drift, ComponentResource coupling, ESC configuration, and Automation API workflows. Triggers on: "Pulumi audit", "stack review", "Automation API review", "ComponentResource design", "ESC audit", "Pulumi secrets", "Pulumi testing".
pareto-optimization
Multi-objective optimization with Pareto frontiers. Use when optimizing multiple conflicting objectives simultaneously, finding trade-off solutions, or computing Pareto-optimal points.
parallel-processing
Parallel processing with joblib for grid search and batch computations. Use when speeding up computationally intensive tasks across multiple CPU cores.
project-sync
Fan out the current repo's CLAUDE.md, .claude/commands, and .claude/skills into Codex, Gemini, and Copilot project-scope analogues (.codex/skills, .gemini/skills, .github/prompts, .github/instructions) plus refresh AGENTS.md, GEMINI.md, and .github/copilot-instructions.md from the project's rule floor. Wraps `dotbabel project-sync`. Use when adopting cross-CLI workflows in a repo for the first time, or when adding a new command/skill that needs to be visible to Codex/Gemini/Copilot. Triggers on: "sync this repo", "project sync", "fan out commands", "wire up gemini for this project", "add this command to codex", "make this skill available in copilot", "regenerate AGENTS.md for this project".
custom-distance-metrics
Define custom distance/similarity metrics for clustering and ML algorithms. Use when working with DBSCAN, sklearn, or scipy distance functions with application-specific metrics.
object_counter
Count occurrences of an object in the image using computer vision algorithm.
pr-conductor
Land one pull request end to end: pre-PR quality gate, open the PR, AI review, apply review fixes, local CI attestation — then STOP with a go/no-go summary. Never merges; merging stays an explicit human call handed to /merge-pr. Understands stacked PRs, where a child must retarget and rebase --onto after its parent squash-merges. Delegates every phase to an existing artifact rather than reimplementing it. On demand only: the default path from a task to a pull request is /deliver. Triggers on: "land this PR", "run the PR pipeline", "run the review pipeline", "ship this branch", "what's blocking my PR".
post-pr-review
Post AI-generated review findings on a GitHub PR as inline review comments via the gh CLI — the bot-style review-comment workflow GitHub Copilot used to provide before it got rate-limited. Idempotent (hidden marker dedup), rate-limit conscious (respects GitHub's 80/min secondary limit), and dry-run by default in interactive mode. Pairs with `review-pr`, which consumes the comments this skill produces and applies fixes. Triggers on: "post PR review", "leave inline comments on PR", "AI review and comment on PR", "Copilot-style PR review".
plan-grader
Grade implementation plans on a 0-10 scale without mutating the workspace. Accepts plan files, quoted inline plans, or best-effort latest-plan discovery for Claude, Codex, Copilot, and Gemini.
fjsp-baseline-repair-with-downtime-and-policy
This skill should be considered when you need to repair an infeasible or non-optimal flexible job scheduling planning schedule into a downtime-feasible, precedence-feasible one while keep no worse policy budget.