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
ssh-ops
Operate remote SSH servers through the bundled CLI and YAML server configurations. Use for remote commands, service restarts, log inspection, file transfer, text edits, deployments, background jobs, and aliases on bash/cmd/powershell servers, even when the user does not explicitly mention SSH. Local-only builds and GitHub releases without a remote server operation are outside this skill's scope. Trigger keywords: deploy, release, ship, server health, SSH, run on server, upload script, download file, server alias, 部署, 上线, 发布, 重启, 查看日志, 服务器健康检查.
phonedriver
Use when the task needs a phone, tablet, iPhone, iPad, Android device, emulator or simulator, or a mobile app: look at the screen, tap, type, scroll, open an app, change a setting, or check what an app shows.
hyper-report
Use this skill when the user asks for a report as a finished document rather than a chat answer: a briefing, business or research report, proposal, status or analysis write-up, or one-page summary. It fixes audience and decision up front, keeps every number and claim traceable to a source, and renders one authored Markdown source into HTML, print PDF, per-page PNG/JPG images, DOCX, or Markdown, with tables and figures where they carry the reader. Do not use it for slide decks, spreadsheet workbooks, single generated images, or reports that only need to live in chat.
hyper-pptx
Use this skill when the user asks to build, edit, or read a PowerPoint deck - create a .pptx from an outline or analysis, add slide transitions or element animations, change text or fonts in a deck someone else made, remove slides, or open a .pptx and report what each slide says. Write a typed JSON spec, render with a bundled engine to a .pptx plus checked slide images, or inspect an existing deck before touching it. Do not use for slide images only, PDF-only output, Google Slides, or Keynote.
hyper-pdf
Use this skill when the user asks to read, extract from, or produce a PDF - text, tables, images, metadata, form fields, OCR of a scan, merge, split, rotate, watermark, encrypt, or a Korean contract, invoice, or report - and wants the result verified rather than assumed. Two generation paths with an explicit routing rule, a zero-dependency macOS runtime, and checks for font embedding, language tag, and structure tree. Do not use for Word or HWP documents, screenshots, slide decks, or numeric charts.
hyper-hwp
Use this skill when the document to read, fill, edit, or produce is a Korean HWP or HWPX file - a .hwp/.hwpx report, 기안문, 공문서, 별지 서식, 신청서 with form fields, or a table that has to come out of one. It drives the rhwp CLI: read text and structure, extract tables and form fields, replace or insert text and table cells, fill fields, insert footnotes, scaffold a new document from a JSON spec, convert HWPX to HWP5, and gate the result with roundtrip and layout checks. Do not use for PDF, DOCX, XLSX, or PPTX files, Hangul word-processor automation through its own COM or UI, or scanned-image OCR.
hyper-excel
Use this skill when the user asks to read, analyze, create, or modify Excel workbooks (.xlsx, .xlsm) or turn CSV data into spreadsheets - inspecting sheets and formulas, editing cells while keeping existing formatting and formulas intact, building formatted reports with charts, or auditing formulas for broken ranges and error values. Do not use for plain dataframes, PDF or slide output, or VBA macro authoring.
hyper-docx
Use this skill when the deliverable is a Word document: create a .docx, edit or restyle one, review it with tracked changes and comments, or check that a generated file opens and lays out as intended. It builds the package from a declarative JSON spec, reads an existing package part by part, applies Korean document conventions (item hierarchy, body size and line spacing, margins, 붙임/끝), records edits as revisions, and verifies the result structurally and visually. Do not use for Google Docs or Hangul (.hwp/.hwpx) files, PDF-only output, spreadsheet or slide formats, or prose that is never rendered into a document.
osmo-phone-calls
AI phone call skill. Use when the user wants an AI to make a real phone call for them — call a business, shop, clinic, restaurant, helpline or a person they know; book, reschedule or cancel an appointment or reservation; reach customer service and get through the phone menu; ask a question and bring back the answer with a summary. Also use to see who called the user's OsmO number, what they wanted, and to change OsmO call screening rules. Free to start in the US, Canada, UK and Australia, where free credits call businesses and toll-free numbers; calling contacts and scheduling need an OsmO app plan. Not for SMS or email.
ui-qa-pass
Runs a branch's app locally and drives a headless browser through the screens the change touches, choosing the states, themes, widths and roles by what the change could break. Takes a screenshot of each, fixes bugs and obvious polish as it goes with before and after shots, and writes a report that leads with the decisions the user needs to make. Use when the user says "QA this", "do a QA pass", "click through every changed screen", "screenshot every state", "test it in a browser and make a report", or before asking for review of a change people can see (app screens, marketing or docs pages, emails). Not for writing or fixing the automated browser test suite.
story-kickoff
Investigates a ticket before any code is written. Has a short path for mechanical changes. Otherwise records the exact code it investigates, checks the ticket's claims against the code, maps callers, existing data and the behavior other code, clients and customers rely on with bounded read-only investigators, then writes a plan whose regression tests each state how they fail on the parent commit and which marks only real judgment calls for the user. Builds test-first, checks the result against every recorded decision before the PR, hands off to the team's review process, and turns what that review found into new lessons. Use when the user says "start work on ENG-123", "pick up this issue", "let's do this ticket", "work on issue 42", "plan this before building it" or "investigate before coding". Not for reviewing a branch or PR that already exists (pr-review or second-opinion-review).
session-wrap-up
Sweeps a working session for loose ends before it closes. Re-reads the saved transcript from disk (Claude Code and Codex formats, including everything before a context compaction), subagent reports, review notes, the PR and the diff, lists every item that was raised but not resolved, checks the tracker for duplicates, and proposes one list with a disposition per item (fix now, ticket with a reason, comment on an existing ticket, save as a rule, drop with a reason). Acts only on what the user approves, then cleans up leftovers and a merged branch's worktree safely. Use when the user says "wrap up", "wrap up the session", "any loose ends?", "anything not in the tracker yet?", "what did we leave unfinished?", or after a feature ships. Not for reviewing a diff for bugs (use your review process).
second-opinion-review
Runs an independent second code review of a committed branch or pull request with another agent's CLI (the Codex CLI is the worked example) unattended in the background. It keeps the second reviewer away from the first reviewer's notes, checks that it left the repository unchanged, then verifies each of its findings against the code before presenting them. Use when the user asks for a second opinion, an independent or impartial review, another model's review, "get Codex to review this", or before pushing a critical fix or a change that needs a careful review. Not for doing your own review of a PR (pr-review) or for fixing failing CI and review comments (fix-ci).
pr-review
Collaborative pull request review with a persistent review folder, independent read-only investigations of bounded focus areas, evidence-backed findings that the user adjudicates one by one, and review comments posted only after the user approves their exact text. Covers someone else's PR and a self-review of your own branch before you open one, and resumes a review after new commits without starting over. Use when the user says "review PR 123", "review this branch", "self-review before I open the PR", "continue the review", "discuss the findings" or "draft the review comments". Not for an unattended review by another model (second-opinion-review) or for fixing failing CI and review comments (fix-ci).
fix-ci
Pushes the current branch, waits for the whole CI pipeline, then fixes every failing job and every valid pull request review comment (from people or review bots) locally as one batch, pushes once, replies to each review thread and resolves the ones its fixes address, and repeats until CI is green with no open threads it can settle. GitHub and GitHub Actions are the worked adapter; other providers need their commands in the settings. Use when the user asks to "fix CI", "get CI green", "get the checks passing", "wait for CI and fix it", "address the review comments" or "resolve the PR comments" on the current branch. Not for reviewing someone else's PR (use your review process) or for an independent second review (second-opinion-review).
check-agent-access
Use when you or the user must choose a tool, API or provider for a task, or someone asks "can my agent use X?": whether an AI agent can sign up for and use each product on its own, from AgentDrives' test drives.
ui-design-canvas
An infinite, Figma-like canvas for UI/UX design work such as mobile apps and web interfaces. It renders images and HTML/CSS/JS pages side by side so designers can review, compare and iterate on designs. Use when the user wants to design screens, build UI prototypes or mockups, or put generated images or HTML pages on a canvas to view them.
workspace-publish
Make bigger changes to a Leeway folder in a workspace (a private fork), show the diff, and publish it back only when the person agrees. Use for multi-file edits, renames or reorganising a Leeway folder, or when asked to fork, draft, stage, review or publish changes on the drive.