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
@microsoft
openclaw
@openclaw
OpenAI
@openai
The Apache Software Foundation
@apache
GitHub
@github
Anthropic
@anthropics
AutoGPT
@Significant-Gravitas
Garry Tan
@garrytan全部 Skills 196587
kv-or-stream
Decide between a KV watch and a JetStream stream for a new communication path. Use when designing communication between components, adding a message flow, or choosing a NATS storage primitive.
handoff
Adapter for handoff.
design
Adapter for design.
review-pr
Review one pull request as an independent, read-only reviewer and post the record comment merge-check reads. The manual trigger for a cross-agent review.
query-pattern
Choose the right query access pattern (GraphQL, MCP, NATS Direct) for a use case. Use when designing query APIs, adding new access points, or choosing between gateway types.
preflight
Select and execute SemTeams Taskfile verification gates for the current diff and record reproducible evidence. Use before implementation pushes, handoffs, or concluding a task.
ponytail
Lazy senior dev discipline: take the smallest complete change that solves the task, protect trust boundaries and tests, track // shortcut: debt, and end conclusions and handoffs with skipped/risk lines. Levels: on, off.
pickup
Resume a repository effort across Claude, Codex, or Gemini by reconciling live git/worktree state, GitHub/Gitea issues/PRs, Hold: items, and the last structured handoff checkpoint.
orchestration-check
Determine whether logic belongs in a reactive rule, workflow, or component. Use when adding orchestration logic, designing multi-step processes, or reviewing boundary violations.
new-payload
Step-by-step checklist for adding a new payload type to the registry. Use when creating new message types for the agentic system or any polymorphic message flow.
kv-or-stream
Decide between KV Watch and JetStream Stream for a new communication path. Use when designing inter-component communication, adding new message flows, or choosing storage primitives.
handoff
Checkpoint current repository work for continuation or transfer across Claude, Codex, and Gemini. Reconciles live git/test/hold state and emits the 5-field pickup prompt.
design
Draft design.md for an OpenSpec change on the design track: file:line context, options with trade-offs, the adopter seam, and owner questions. Read-only on code; drafts and never rules or implements.
skypilot-multi-cloud-orchestration
Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs across providers.
skypilot-multi-cloud-orchestration
Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs across providers.
modal
Cloud computing platform for running Python on GPUs and serverless infrastructure. Use when deploying AI/ML models, running GPU-accelerated workloads, serving web endpoints, scheduling batch jobs, or scaling Python code to the cloud. Use this skill whenever the user mentions Modal, serverless GPU compute, deploying ML models to the cloud, serving inference endpoints, running batch processing in the cloud, or needs to scale Python workloads beyond their local machine. Also use when the user wants to run code on H100s, A100s, or other cloud GPUs, or needs to create a web API for a model.
lambda-labs-gpu-cloud
Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.
lambda-labs-gpu-cloud
Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent filesystems, or high-performance multi-node clusters for large-scale training.