Skills
97,222 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Creates, edits, or improves agent skills (SKILL.md). Use whenever the task is to author a new skill, refactor or shorten an existing one, fix its frontmatter or triggering description, or split it for progressive disclosure. Load this before researching other skills or writing any SKILL.md.
Use when an agent needs to delegate a task to the OpenAI Codex CLI from another agent environment such as Claude Code, OpenClaw, or similar. Covers checking whether Codex CLI is installed, running one-off Codex prompts with `codex exec`, resuming sessions, collecting outputs, attaching images or files as input with `-i`/stdin, and handling Codex image generation including finding and reporting generated image file paths.
End-to-end LLM Observability pipeline for an instrumented ml_app — classify production traces, root-cause failures, bootstrap evaluators, then (optionally) sample + publish a dataset, generate + run an experiment, and analyze results. Six narrated phases with a standardized banner and a “continue” checkpoint between each. Pure orchestration over the dd-llmo sub-skills (`llm-obs-session-classify`, `llm-obs-trace-rca`, `llm-obs-eval-bootstrap`, `llm-obs-experiment-py-bootstrap`, `llm-obs-experiment-analyzer`). Use when user says “run the eval pipeline”, “go from traces to evals”, “bootstrap evals end to end”, “classify then RCA then bootstrap”, “build an eval set from scratch”, “onboard me to datasets and experiments”, “walk me through experiments”, “I have an ml_app, now what”, “LLM Obs onboarding”, “guided experiment setup”, “from traces to experiments”, or wants a deterministic, narrated tour from production data through evaluators, datasets, and experiments. Stop early with `--stop-after <phase>` to short-circuit at evaluators or dataset, or resume mid-flow with `--start-at <phase>`
Plan, create, review, and maintain AI agents, model workflows, skills, and durable project context. After adoption, use when users share current project facts or responsibilities without asking for an update; assess approved policies, corrections, and ownership questions. Ordinary code needs a declared relationship; stay silent on a miss. Initialize only when asked.
Slice an epic (with its architecture decisions) into PIV-sized tickets with a dependency graph, then create them in your tracker (Jira via the Atlassian MCP, or GitHub/Linear/local). Accepts the epic and its architecture as one doc or as an epic plus a separate linked architecture page. Turns a large strategic doc into the discrete units of work that the PIV loop consumes.
Full lifecycle project management for independent developers: validate demand, choose keywords and products, build a website or native app, launch and integrate, acquire users through SEO/GEO, monetize with payments, monitor and iterate, and hand off across sessions. Use when the user mentions rankup or /rankup (check, review, init; doctor only for organizing project .rankup records); when the current directory or workspace contains .rankup/, or they need to read, continue, or organize project plans, roadmaps, tasks, PRDs, handoffs, or progress documents; when they say “continue,” “pick up where we left off,” “where did we leave off,” or “what’s next” in the context of these project records; and for project planning, demand validation, keyword research and SERPs, standalone Google Trends queries or comparisons, AI search recommendations (GEO), website audits, sitemap/IndexNow/Search Console, traffic, website building, launches, payments (Stripe, Anyway, PayPal), monetization, and growth. SEO/GEO are primary methods, not the boundaries of applicability; non-SEO project plans and native macOS, iOS, and iPad apps are also covered. Do not activate for copywriting alone (including topic writing such as SEO trend blogs), visual design alone, general development unrelated to project management, or infrastructure troubleshooting (including Cloudflare deployment errors and ordinary CLI doctor); if .rankup/ exists in the directory, only add the obligation to maintain project records. Delegate backlink execution to backlink, browser automation to opencli, image generation to imagegen, and multi-model task dispatch to agent-fleet.
Current Claude AI and Claude Code capabilities, features, and limits. Consult before making claims about what Claude can or cannot do, what features exist, model availability, pricing, or environment differences. Use when answering “Can Claude do X?”, comparing Claude AI vs. Code, checking current model specs, or advising on which environment suits a task.
Compact the current conversation into a large standalone handoff file on disk, then open a BRAND NEW Claude Code session named after the current chat with an incremented counter, seeded with only a pointer to that file, leaving the current chat fully intact. Use this whenever the user wants a clean context window without losing what was learned, or says any of — "compact to a new chat", "compact but start fresh", "carry this over to a new session", "hand this off", "new chat with this context", "my context is getting heavy", "this chat is too long", "start over but remember this", "dump the context and open a fresh one", or asks for a handoff, a context dump, or a session summary they can reuse later. Also use it when a long investigation is finished and the follow-up work deserves its own window. Prefer this over plain /compact whenever the user wants the old chat to stay readable afterward, and reach for it even if they never say the word "compact".
Orchestrates debates among multiple advisors on high-impact architecture, technology, or product decisions. Dispatches 3–5 domain-archetype subagents (pragmatic-engineer, architect-advisor, security-advocate, product-mind, devils-advocate, the-thinker) through opening statements, tensions, position evolution, and synthesis phases. Preserves dissent and delivers actionable recommendations with documented risks. Use when evaluating trade-offs, stress-testing a PRD or tech spec, resolving dilemmas with multiple viable options, or when a decision needs diverse expert perspectives. Don’t use for simple yes/no questions, factual lookups, creative brainstorming without trade-offs, or tasks where one expert perspective is enough.
Token-efficient persistent memory system for Claude Code that saves ~67% tokens on session warm-up (verified with tiktoken). Layered architecture with progressive loading, compact encoding, branch-aware context, smart compression, session diffing, conflict detection, session continuation protocol, and recovery mode. Activates at session start (if MEMORY.md exists), on "remember this", "pick up where we left off", "what were we doing", "wrap up", "save progress", "don't forget", "switch context", "hand off", "memory health", "save state", "continue where I left off", "context budget", "how much context left", or any session start on a project with existing memory files. This skill solves two problems at once: Claude forgetting everything between sessions, AND sessions hitting context limits too fast. It replaces thousands of wasted re-explanation tokens with a compact, structured memory load that gives Claude full project context in under 2,000 tokens.
Design ObjectStack AI skills, tools, knowledge sources, and the open-edition MCP server surface. Use when the user is adding `*.skill.ts` / `*.tool.ts`, configuring an LLM provider, wiring the tools a skill grants, or indexing ObjectStack data as a knowledge source for RAG. Agents themselves are platform-internal (`ask` / `build`) — third parties extend them via skills, not by authoring `*.agent.ts`. Do not use for general LLM prompting questions unrelated to ObjectStack metadata.
Step-by-step guide for creating your own Claude Skills, from deciding whether a skill is the right tool to writing the SKILL.md file, structuring reference material, and making it trigger reliably. Use when you want to package a workflow, framework, or repeated task into a reusable Skill, when an existing skill is not triggering or not loading the right context, when you are auditing a skill that is underperforming, or when you want to publish a skill for others. Also triggers when someone asks "how do I make a skill" or "what makes a good skill". Useful for individuals, teams, and anyone publishing skills publicly.
Discover and follow fullstackrecipes (setup guides, skills, cookbooks) by fetching their Markdown. The meta-skill for finding and applying recipes correctly. Use when planning a build, adding a tool, or deciding which recipe or skill to follow.
Take a PRD and build a dark factory around it: a repository that takes work in as an issue and ships validated code out with nobody at the keyboard, one component at a time, in the user’s actual repo. Covers five components in construction order: the guidance layer, validation harness, workflow-driven repo, deployment, and the trigger that makes it autonomous. It is agnostic about which coding agent runs underneath (Claude Code, Codex, Archon, the Agent SDK, Cline, Goose, Amp, Pi). It encodes the AI coding process the user already runs rather than replacing it. Requires a PRD as input and deliberately does not write one. Use when the user wants to build a dark factory, an autonomous or self-driving repository, a software factory, an agent that ships its own code, an unattended or overnight coding loop, asks how to reach level 4 or 5 of AI coding autonomy, or mentions dark factory, lights-out coding, autonomous PRs, or a repo that maintains itself.
Register AI agents on Ethereum mainnet using ERC-8004 (Trustless Agents). Use when the user wants to register their agent identity on-chain, create an agent profile, claim an agent NFT, set up agent reputation, or make their agent discoverable. Handles bridging ETH to mainnet, IPFS upload, and on-chain registration.
Manage the lifecycle of ExecPlans — self-contained, living specifications for multi-step work. Creates plans in the correct format, enforces living section updates, and handles the active → completed transition. Use for any work expected to take more than one session or touching more than 3 files. Triggers: "create a plan", "write a plan", "start plan", "continue plan", "resume plan", "finish plan", "complete plan", multi-step features, refactors, or tasks spanning sessions.
Cross-model second opinion from Google Gemini — a different AI reviewing the same changes, with deep Google ecosystem knowledge. Three modes: review (pass/fail gate for Google Ads campaigns, SEO metadata, or code), challenge (adversarial stress-test that tries to break your changes), and consult (open Q&A with Gemini on Google Ads strategy, SEO best practices, or implementation questions). Use when the user says "gemini review", "ask gemini", "gemini challenge", "second opinion from gemini", "consult gemini", "stress test with gemini", "what would gemini say", "cross-model review", or "get another opinion". Voice aliases: "gem", "gemini check". Especially useful for Google Ads changes, SEO metadata updates, campaign structure decisions, keyword strategies, and bid/budget changes — Gemini has native Google ecosystem knowledge that complements Claude's analysis.
Use when the user explicitly wants autonomous progress without routine clarification stops: "plow ahead", "do not stop", "use your best judgment", "keep going until done", "finish while I am away", "do not ask questions unless truly blocked", or similar. Convert ordinary ambiguity into stated assumptions, proceed through implementation and validation, stop only for true blockers, and end with a clear recap of decisions, changes, verification, and residual risk.
Primes the agent with focused understanding of the backend portion of the codebase — API routes, services, data models, and database layer — without loading unrelated frontend code. Use at the start of a session when the work is scoped to API endpoints, business logic, or data access. Optionally pulls external task context from Jira issues and Confluence pages first.
Close out (or revisit) a Claude Managed Agent build — refresh the overview page, recap every primitive the founder now owns, show the run log and live status, suggest 1–2 tailored next upgrades, and sweep hygiene (sessions archived, key only in .env, no literal dates in deployment kickoffs). Use when the founder says "/wrap-up", "wrap up", "close it out", "where do things stand with my agent", or at the end of a /launch-your-agent build.
Use when users say "create a skill", "make a new skill", "build a skill", "skill for X", "audit this skill", "review this skill", "check skill quality", "fix this skill", "improve this skill", "refactor this skill", "update this skill", or "optimize this skill", or when creating, refactoring, or auditing domain expertise into agent skills with specialized knowledge, workflows, or tool integrations.
Use this skill to explain or reason about the foundational concepts of context engineering: what context is, the anatomy of a context window, how attention mechanics work, the U-shaped attention curve, why context quality matters more than quantity, and the mental models needed to interpret every other context-engineering decision. Use it for conceptual explanation, onboarding, and background reading. Route operational work to the specialized skills: debugging attention failures goes to context-degradation, token-efficiency work to context-optimization, conversation summarization to context-compression, and project-shape decisions to project-development.
Manage AI agent memory with a P0/P1/P2 priority system and automatic archiving. Use when setting up memory management, cleaning up MEMORY.md, creating lessons files, configuring automatic archive cron jobs, or reviewing memory structure. Reduces token usage by 70–80% while maintaining recall through semantic search.
Builds and improves Browserbase Agent API demos through an Autobrowse-style outer loop: run a fixed task, collect Agent messages and session logs, score the result, revise one system-prompt heuristic, and confirm convergence. Use when creating a Browserbase Agents demo or POC, optimizing an Agent system prompt, diagnosing flaky Agent runs, or applying auto-research/autobrowse to the Browserbase Agents API.
Use this skill when the user asks to “create a plugin,” “scaffold a plugin,” “understand plugin structure,” “organize plugin components,” “set up plugin.json,” “use ${CLAUDE_PLUGIN_ROOT},” “add commands/agents/skills/hooks,” “configure auto-discovery,” or needs guidance on plugin directory layout, manifest configuration, component organization, file naming conventions, or Claude Code plugin architecture best practices.
How to work in a Software Lifecycle project (the `software-lifecycle` starter pack): proposals → decisions → specs → postmortems, plus guides. Read when the project has these folders, or when asked how this project is organized. Carries the doc lifecycle, status flows, and per-folder agent behaviors so that guidance does not live inside template bodies or folder descriptions. The five workflows — frame a proposal, write a spec, record a decision, write a postmortem, review a design — each ship as their own sibling skill in this pack. Complements the platform `open-knowledge` skill; does not replace it
Measure whether a repository's AI instructions still earn their place, by running the same real task many times with the layer intact and with it stripped, then grading every rule against what actually changed. Runs both arms itself in throwaway git worktrees and never touches the working tree. Agent-agnostic across CLAUDE.md, AGENTS.md, .claude/, .agents/, .cursor/rules, .clinerules, .windsurfrules and copilot-instructions. Use when the user wants to prune, audit, clean up, shrink or "delete" their CLAUDE.md, AGENTS.md, cursor rules, agent instructions or AI layer; when they ask whether their rules are still needed, whether their context is bloated, or what to cut; or when they mention ablating, ablation, or testing their agent without its instructions.
Automates browser interactions for web testing, form filling, screenshots, and data extraction. Use when the user needs to navigate websites, interact with web pages, fill forms, take screenshots, test web applications, or extract information from web pages.
Scan how you actually work with your coding agent and surface what to encode next. Point it at ONE run’s artifacts to find what would have prevented a specific failure (the reactive loop — “that went wrong, what should change in the AI layer?”), or at a window of session logs to find recurring patterns worth building (the proactive scan). Agent-agnostic. Outputs a shape-only HTML report. Use to evolve your system from real usage.
Mine past Claude Code and Codex conversations for uncaptured patterns, corrections, and knowledge. Cross-references with existing brain content. Triggers: "ruminate", "mine my history", "what have I been working on", "review past sessions", "extract learnings".
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.