agent-skills

Plugin agent-skills in joeblackwaslike/agent-skills. Commands read from the repo's manifest — yesterday.

Skills (46)

• agent-instructions

  Use when creating or editing CLAUDE.md, AGENTS.md, GEMINI.md, a SKILL.md, a runbook, or any file that instructs an AI coding agent: "set up Claude for this project", "what goes in my AGENTS.md", "add a rule", "edit this skill". Covers scope, content taxonomy, multi-tool compatibility, RED-GREEN edit

• authoring-agent-skills

  Use when creating or maintaining a skill in the agent-skills repo, especially a working-with-X doc-wrapper: naming, SKILL.md frontmatter and descriptions, doc-fetching scripts, freshness metadata, registration. Triggers: "create a skill", "add a working-with-X skill", "add a skill to this repo". Not

• autonomous-agent-operations

  Use when operating solo on a task handed off without the user available for interactive check-ins — ask every clarifying question up front, decide and file a labeled bd ticket on a genuine mid-run fork rather than blocking or guessing silently, and summarize plus record a worked example at the end.

• beads-operations

  Use when running, maintaining, or troubleshooting beads (`bd`) day-to-day — deciding bd vs TodoWrite, Joe's shared-server conventions (`bd init --shared-server --skip-agents`, the shared Dolt server on port 3308), the ready→claim→note→close workflow, compaction recovery, recovering from broken/nasty

• best-practices-for-agentic-development

  Use when designing or reviewing an agent system: agent loops, tool interfaces, MCP server design, subagent architecture, or skill-writing methodology (RED/GREEN pressure tests). Not for plugin packaging (developing-*-plugins) or CLAUDE.md/AGENTS.md files (agent-instructions). last_updated: "2026-08-

• code-architectures

  Use when choosing or structuring how a system is built: hexagonal layering, CQRS, event sourcing, event-driven design, sagas, microservices vs monolith, strangler-fig migration, service timeouts and contracts, CAP and eventual consistency. Not for aggregates or bounded contexts (domain-driven-design

• design-patterns

  Use when picking or implementing a named pattern: Strategy, Observer, Adapter, Decorator, Factory, Repository, Unit of Work, Dependency injection, Retry with backoff, Circuit Breaker, Idempotency key, Outbox, Result type. Not for whether to abstract at all (use engineering-principles), SOLID refacto

• devcontainers

  Use when building, modifying, using, or distributing Dev Containers — creating or editing devcontainer.json from scratch, adding devcontainer features, writing Dockerfiles for dev containers, authoring custom features or templates, publishing to GHCR, using the devcontainer CLI (devcontainer up, bui

• developing-claude-code-plugins

  Use when building a Claude Code plugin: plugin.json, hooks.json, skills/, commands/, agents/, .mcp.json, userConfig, ${CLAUDE_PLUGIN_ROOT}, local testing with /reload-plugins, or releasing. Hook, command, agent, MCP, and settings authoring references included. Not for official CLI docs (working-with

• developing-codex-plugins

  Use when building a Codex CLI plugin: .codex-plugin/plugin.json, marketplace.json, skills, hooks, MCP servers, local testing, or releasing. Not for Codex CLI usage docs (working-with-codex) or sharing one plugin across tools (multi-provider-plugins). last_updated: "2026-06-04

• developing-cursor-plugins

  Use when building a Cursor plugin: manifest, rules (.cursor/rules/*.mdc), skills, commands, hooks, MCP servers, local testing, or releasing. Not for Cursor IDE usage docs (working-with-cursor) or sharing one plugin across tools (multi-provider-plugins). last_updated: "2026-06-05

• developing-for-github

  Use when building software that integrates with GitHub — GitHub Apps, OAuth apps, webhook consumers, bots, CI checks, and automation built on the GitHub API. Covers the GitHub App auth model (private key → JWT → installation access token), choosing GitHub App vs OAuth app vs PAT, app permissions and

• developing-gemini-plugins

  Use when building a Gemini CLI extension or Antigravity CLI plugin: gemini-extension.json, GEMINI.md context, skills, commands, hooks, MCP servers, `gemini extensions link`, or releasing. Not for Gemini CLI usage docs (working-with-gemini) or sharing one plugin across tools (multi-provider-plugins).

• developing-opencode-plugins

  Use when building an OpenCode plugin or skill: plugin code, skills.paths in opencode.json, hooks, local testing, or releasing. Not for OpenCode usage docs (working-with-opencode) or sharing one plugin across tools (multi-provider-plugins). last_updated: "2026-06-04

• docusaurus-docs-builder

  Use when creating, redesigning, rewriting, or improving a Docusaurus documentation site, docs portal, developer docs, product docs, API docs, guide site, knowledge base, or documentation landing experience, especially when the user wants engaging, attractive, polished, conversion-aware, visual, MDX-

• domain-driven-design

  Use when modeling a business domain: aggregates, entities, value objects, domain events, repositories, bounded contexts, ubiquitous language. Also when an agent builds a non-trivial domain with real invariants. Not for layering, CQRS, event sourcing, sagas, event-driven design or services (use code-

• engineering-principles

  Use when writing, designing, refactoring or extending code: where logic belongs, how much abstraction is enough, module and interface structure, error handling, validation at boundaries, naming, what a test should prove. Triggers on "how should I structure this", "is this too complex", "should I abs

• git-github-workflows

  Use when committing, opening a PR, driving it to merge, answering reviewer or AI-bot feedback, working in a worktree, cutting a release, or cleaning up branches. Not for command flags (git worktree, gh pr: working-with-git, working-with-github), CI workflow YAML (working-with-github-actions), GitHub

• github-readme-overhaul

  Use when writing, rewriting, branding, visually overhauling, or optimizing a GitHub README.md for installs, downloads, stars, marketplace conversion, OSS adoption, or immediate product value, including badges, hero graphics, tagline, pain-point messaging, quickstart, terminal demos, screenshots, GIF

• interactive-system-docs

  Build a self-contained interactive HTML visualization that explains how a complex system works — architecture, data flows, state machines, processing pipelines, and search patterns — with D3.js graphs where every node is clickable, all rendered from a single file with no build step. Use when the use

• multi-provider-plugins

  Use when porting a Claude Code plugin to Codex, OpenCode, Cursor, or Gemini CLI: provider adapters, bootstrap patterns, tool-name mapping, or a shared skills library. Not for single-platform plugin work (developing-<platform>-plugins). last_updated: "2026-06-04

• recovering-session-context

  Reconstruct what happened in prior agent sessions from their transcripts and surface everything raised but never resolved — unanswered questions, work promised and not done, topics dropped, partial completions, and incomplete todos. Use when resuming after an interruption or compaction ("pick up whe

• reviewing-code

  Use before opening a PR or sharing finished work, to self-review your own diff as a reviewer who didn't write it. Triggers on "/self-review", "about to open a PR", "polish this before I share it", "is my diff ready", or before `gh pr create`. Not for reviewing someone else's PR. Not for writing or d

• senior-engineering-best-practices

  Use before proposing a workaround for third-party code, answering from memory about what a library supports, patching the same area a second time, adding a dependency, or taking an action whose blast radius is unconsidered. Routes to the matching runbook. For TDD, SOLID, DDD, debugging, or code revi

• solid-principles

  Use when designing or refactoring classes, modules, or interfaces and the user mentions SOLID, SRP, OCP, LSP, ISP, DIP or dependency injection, or code shows a god class, a growing type-tag switch, a subclass that throws "not supported", or a fat interface. TypeScript and Python examples. Skip throw

• test-driven-development

  Use when implementing any feature, fixing any bug, or changing behavior in TypeScript or Python — and whenever the user mentions TDD, test-first, test-driven, red-green-refactor, "write tests", "add tests", unit/integration tests, pytest, vitest, jest, mocking/test doubles, coverage gates, or retrof

• vscode-extension-builder-lawvable

  Use when building, converting, testing, debugging, packaging, publishing, or reviewing VS Code extensions, including commands, webviews, custom editors, tree views, language features, AI/Language Model API integrations, Language Model Tools, extension host behavior, and .vsix distribution.

• web

  Use before WebSearch or WebFetch, and for any web search, URL read, library-docs lookup, or multi-source research. Also when a fetch is blocked (403, 429, captcha, login wall, empty SPA). Routes to SearXNG, Exa, Context7, Ref, Firecrawl, Parallel, or stealth-browser. Not for multi-step browser autom

• working-with-bash

  Use when working with Bash: commands, builtins, shopt/set -o, parameter expansion, [[ ]] vs [ ], globs, redirection, .bashrc/.inputrc, strict mode (set -euo pipefail), quoting bugs, POSIX portability, completion, Readline, or ShellCheck/shfmt/bats. Version-exact manual bundled. For Zsh use working-w

• working-with-beads

  Use when working with the beads CLI (`bd`) — issue/task tracking, dependencies, the full command surface (create, ready, dep, epic, gate, swarm, federation, dolt, worktree, compaction), `bd prime`, and the beads data model. Provides an offline CLI reference generated verbatim from the pinned `bd` bi

• working-with-claude-code

  Use when you need the official Claude Code docs: CLI flags, settings.json, hooks reference, MCP config, skills, subagents, slash commands, plugin and marketplace reference, permissions, or the weekly changelog. Not for building a plugin step by step (developing-claude-code-plugins) or editing CLAUDE

• working-with-codex

  Use when you need the official OpenAI Codex CLI docs: configuration, AGENTS.md, skills, hooks, plugins, MCP. Not for building a Codex plugin (developing-codex-plugins). last_updated: "2026-06-04

• working-with-cursor

  Use when you need the official Cursor docs: rules (.cursor/rules/*.mdc), skills, plugins, MCP servers, agents, commands, hooks, configuration. Not for building a Cursor plugin (developing-cursor-plugins). last_updated: "2026-06-05

• working-with-dolt

  Use when working with Dolt — the version-controlled SQL database that branches, diffs, merges, and clones like Git while speaking the MySQL wire protocol ("Git for data"). Covers the full `dolt` CLI (init, sql, sql-server, branch, merge, diff, commit, remote, backup, gc, fsck, ci, profile, creds, co

• working-with-gemini

  Use when you need the official Gemini CLI or Antigravity CLI docs: extensions, skills, hooks, GEMINI.md, configuration, models. Not for building an extension (developing-gemini-plugins). last_updated: "2026-06-04

• working-with-git

  Use when you need to know how Git itself works: command flags (including worktree and branch), concepts, internals, Pro Git, and guides for rewriting history, bisect, recovering lost commits, conflicts, Git servers. Not for how we use git (commits, PRs, worktrees, review loops: git-github-workflows)

• working-with-github-actions

  Use when building, configuring, debugging, or optimizing GitHub Actions CI/CD workflows. Covers workflow YAML syntax, event triggers, job orchestration, matrix builds, reusable workflows, composite actions, secrets and variables, artifact and cache management, OIDC-based cloud deployments, self-host

• working-with-github

  Use when you need to know how GitHub the platform works: gh commands and flags, REST and GraphQL, rate limits, tokens, rulesets, issues, release endpoints, Dependabot and CodeQL config. Not for how we drive PRs, reviews and releases (git-github-workflows), raw git (working-with-git), workflow YAML a

• working-with-opencode

  Use when you need the official OpenCode docs: CLI, opencode.json configuration, plugins, skills, hooks, agents. Not for building an OpenCode plugin (developing-opencode-plugins). last_updated: "2026-06-04

• working-with-pieces

  Use when working with Pieces for Developers (Pieces OS, Desktop, CLI, IDE plugins, the Pieces MCP server, long-term memory) or when asked "what was I working on", "what did I do yesterday/last week", or "what did I have open about X". Covers MCP setup, LTM queries, snippet search, and the Pieces SDK

• working-with-release-please

  Use when configuring or debugging release-please: release-please-config.json, .release-please-manifest.json, release types, monorepo components, the release-please-action workflow, or "why didn't a release PR appear". Not for raw gh release commands (working-with-github), general workflow YAML (work

• working-with-serena

  MANDATORY before reading or editing any code file. Use when working with Serena — the MCP toolkit (github.com/oraios/serena) that wraps Language Servers to give symbol-level code retrieval and editing — or any time you are about to Read/Edit/Write a code OR config/data file, Grep/Glob for a symbol,

• working-with-vercel-ai-sdk

  Use when working with the Vercel AI SDK (the `ai` package) — generateText, streamText, generateObject, streamObject, embeddings, tools/tool-calling, agents (ToolLoopAgent), AI SDK UI hooks (useChat, useCompletion), RSC, providers (OpenAI/Anthropic/Google/AI Gateway), and migrations. Provides compreh

• working-with-vercel-api

  Use when working with the Vercel **REST API** and **@vercel/sdk** — the programmatic interface to the Vercel hosting platform at api.vercel.com. Covers bearer-token auth (VERCEL_TOKEN), team scoping (teamId/slug), per-endpoint versioning (/v1, /v8, /v9), pagination, errors/rate limits, and the typed

• working-with-vercel

  Use when working with the Vercel **hosting platform** — the `vercel`/`vc` CLI (deploy, dev, build, env, dns, domains, blob, edge-config, firewall, rolling-release, logs, link, pull, tokens, redeploy, rollback, promote, mcp, …), the official hosted **MCP server** (https://mcp.vercel.com — OAuth; tool

• working-with-zsh

  Use when working with Zsh — any zsh command, builtin, option (setopt), glob qualifier, parameter/special variable, or expansion lookup; Bash→Zsh migration and behavioral differences; .zshrc/.zshenv/startup-file configuration; the completion system (compinit/zstyle) and ZLE/bindkey line editing; fram

Slash commands (4)

• /agent-skills:commit-push-pr

  Verify, commit, self-review, push, and open a PR from the repo's template, then hand it to /pr-loop

• /agent-skills:commit

  Stage and commit changes as Conventional Commits, splitting into atomic commits when warranted

• /agent-skills:git-cleanup

  Remove landed worktrees and merged local branches, prune stale refs

• /agent-skills:self-review

  Review your own diff with fresh-eyes reviewers, fix the findings, and report whether it is ready

Similar plugins