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 196838
version-sdlc
Use this skill when bumping a project version, creating a git release tag, generating a changelog, or performing a full semantic release workflow, updating an existing changelog entry for the current version, or retagging the current version at HEAD. Consumes pre-computed context from skill/version.js and handles the complete release process. Use --changelog without a bump type to update the changelog for the already-tagged current version. Use --retag to move an existing tag to HEAD. Arguments: [major|minor|patch|<label>] [--init] [--pre <label>] [--no-push] [--changelog] [--hotfix] [--retag] [--auto]. The positional `<label>` form (e.g. `version-sdlc rc`) is sugar for `--bump patch --pre <label>` and accepts any pre-release label matching `^[a-z][a-z0-9]*$`. Triggers on: version bump, create release, bump version, tag release, generate changelog, semantic versioning, semver bump, pre-release, release candidate, retag release. Use --auto to skip interactive approval prompts (release plan is still displayed).
verify-pipeline-sdlc
Use this skill to analyze a failed CI run on a PR and either apply a minimal fix or emit a proposal. Dispatched automatically by ship-sdlc's verify-pipeline step under --auto, or invoked standalone via /verify-pipeline-sdlc --pr <N>. Triggers on: analyze CI failure, fix failing checks, post-PR CI verification, verify-pipeline.
ship-sdlc
Use this skill when shipping a feature end-to-end after plan acceptance: executing, committing, reviewing, fixing critical issues, versioning, and opening a PR in one flow. Dispatches every sub-skill (including execute-plan-sdlc) as an Agent for context isolation, with structured return values driving the pipeline state machine. Arguments: [--auto] [--steps <csv>] [--quick] [--quality full|balanced|minimal] [--bump patch|minor|major|<label>] [--draft] [--dry-run] [--resume] [--plan <path>] [--init-config]. The `<label>` form for --bump (e.g. `--bump rc`) is forwarded to version-sdlc, where it is interpreted as `--bump patch --pre <label>`; labels must match `^[a-z][a-z0-9]*$`. Triggers on: ship it, ship this, full pipeline, execute to PR, ship feature, run the whole thing.
setup-sdlc
Use this skill when setting up the SDLC plugin for a project, initializing configuration, or when any skill reports missing config. Renders a selective-section menu so users choose which sections to configure; each selected section prints a verbose header (purpose, files-modified, consuming skills, per-option description) before any prompt. Supports direct sub-flow entry via --only, --dimensions, --pr-template, --guardrails, --execution-guardrails, --openspec-enrich, --plan-template. Arguments: [--migrate] [--skip <section>] [--force] [--only <ids>] [--dimensions] [--pr-template] [--guardrails] [--execution-guardrails] [--openspec-enrich] [--remove-openspec] [--plan-template] [--add] [--no-copilot]
review-sdlc
Use this skill when reviewing code changes across project-defined dimensions (security, performance, docs, concurrency, etc.). Runs skill/review.js to pre-compute all git data, then delegates to the review-orchestrator agent. Arguments: [--base <branch>] [--committed] [--staged] [--working] [--worktree] [--set-default] [--dimensions <name,...>] [--dry-run]. Triggers on: review changes, code review, review PR, multi-dimension review, run review.
received-review-sdlc
Use this skill when responding to code review feedback on a pull request or inline reviewer comments. Covers reading, verifying, evaluating, and responding to reviewer comments with a dual self-critique gate — prevents performative agreement and ensures technical rigor. Can be launched manually or automatically after /review-sdlc. Triggers on: process review feedback, respond to review, handle review comments, address PR feedback, fix review findings, received-review.
pr-sdlc
Use this skill when creating or updating a pull request, updating a PR description, or generating PR content from commits and diffs. Handles the full PR workflow: consumes pre-computed context from skill/pr.js, generates description with plan-critique-improve-do-critique-improve, user review, and gh CLI execution. Auto-labels PRs based on context signals (branch, commits, diff, Jira) with mandatory approval. Arguments: [--draft] [--update] [--base <branch>] [--auto] [--label <name>]. Use --auto to skip interactive approval. Triggers on: create PR, open pull request, update PR, write PR description, PR summary, describe changes for a pull request.
plan-sdlc
Use when writing an implementation plan from requirements, a spec, a design doc, or a user description. ALWAYS use when plan mode is active — this is the designated plan-mode skill. Analyzes scope, maps file structure, decomposes into classified tasks with dependencies, and produces a plan ready for execute-plan-sdlc. Triggers on: write plan, create plan, plan this, break this into tasks, implementation plan, plan mode.
jira-sdlc
Use this skill when creating, editing, reading, viewing, searching, transitioning, commenting on, or linking Jira issues using Atlassian MCP tools. Caches project metadata (custom fields, workflows, transitions, user mappings) to eliminate redundant discovery calls. Supports multi-project repos via jira.projects, and skipping workflow discovery for CI. Arguments: [--project <KEY>] [--force-refresh] [--init-templates] [--site <host>] [--skip-workflow-discovery]. Triggers on: create jira issue, edit jira ticket, search jira, transition jira, jira comment, link jira, assign jira, log work jira, bulk jira operations, manage jira, jira template, read jira, view jira, show jira, get jira, fetch jira, jira details, add comment, comment on jira, reply to jira, jira ticket, jira issue.
harden-sdlc
Use this skill after an SDLC pipeline failure to analyze hardening surfaces (plan and execute guardrails, review dimensions, copilot instructions) and propose user-approved edits that would prevent the same class of failure next time. Strengthen-only in v1 — never relaxes or removes existing rules. Required arguments: --failure-text <string> --skill <caller-name>. Optional: --step, --operation, --exit-code, --error-type, --user-intent, --args-string. Triggers on: harden, strengthen guardrails, prevent this failure, learn from this failure, after pipeline failure.
execute-plan-sdlc
Use when the user wants to execute an implementation plan with adaptive intelligence — classifies tasks by complexity and risk, builds optimized dependency waves, critiques wave structure before dispatch, verifies results after each wave, and recovers from failures without stopping. Self-contained: no external sub-skills required. Triggers on: execute plan, run plan, implement plan, autonomous execution, execute this plan. Requires an explicit plan file — a positional path or `--plan <path>` — at invocation (or a resumable state file, see `--resume`); it is never inferred from conversation context, even when a plan was just discussed or accepted in this session (R41).
error-report-sdlc
Internal skill invoked by other SDLC skills when they encounter an actionable error (script crash, CLI failure, persistent API error, build failure after retries). Proposes creating a GitHub issue in rnagrodzki/sdlc-marketplace to track the error with full context capture, two-gate user consent, and pre-flight verification. NOT user-invocable — only dispatched from within another skill's error handling path. When dispatched, follow ./REFERENCE.md for the full procedure.
commit-sdlc
Use this skill when committing staged changes, creating a git commit, or generating a commit message. Analyzes staged diff and recent commit history to generate a message matching the project's style. Stashes unstaged changes to isolate the commit, commits after user confirmation, and auto-restores the stash. Arguments: [--no-stash] [--scope <scope>] [--type <type>] [--amend] [--auto] [--force-default-branch]. Use --auto to skip interactive approval. Triggers on: commit changes, create commit, write commit message, git commit, smart commit, commit staged, stage and commit.
validate-skill-tests
Use after creating or modifying a plugin skill to verify promptfoo test coverage. Checks dataset existence, fixture reference validity, new behavior coverage, and assertion quality.
validate-skill-docs
Use after modifying plugin skills, docs/skills/ pages, or site/src/data/skills-meta.ts. Validates content consistency between SKILL.md files and their documentation surfaces: skills-meta.ts field accuracy, connection validity, doc template compliance, and flag documentation.
validate-plugin-consistency
Use after modifying any file under plugins/sdlc-utilities/ (skills/*/SKILL.md, scripts/*.js). Runs the consistency validation script to catch structural issues before they reach users: wrong script resolution order, skills missing prepare-script execution, missing mktemp or exit-code handling, missing frontmatter fields, missing user-invocable flags, missing docs/skills/ documentation, missing skills-meta.ts entries, missing README table rows, temp file cleanup gaps.
test-report
Use this skill when analyzing promptfoo test results and generating evidence reports. Consumes pre-computed data from test-report-prepare.js, generates categorized markdown report with failure diagnosis and fix plan. Triggers on: test report, analyze test results, promptfoo report, test evidence, test failures, generate evidence.
promptfoo-results
Query Promptfoo behavioral test results from the local SQLite database. Use when asked about test results, eval outcomes, failures, what passed or failed, test comparisons, or promptfoo reports.