Skills
97,725 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
How to explore and make sense of PostHog Signals scouts — the scheduled agents that scan a project and write reports into the Signals inbox. Use when a user wants to understand what scouts they have, how each one is behaving, and whether the fleet is actually working. Covers surveying the fleet and its schedules, reading recent scout runs and drilling into a single run's reasoning, inspecting the durable scratchpad memory the fleet has built up, tracing a run to the reports it wrote or edited, reading the follow-up checks still measuring a report, and assessing a scout's health and performance over time (cadence, success rate, report rate, signal-to-noise). Read-only and exploratory — to write or tune a scout, use `authoring-scouts` instead. Trigger on "what are my scouts doing", "how is my <x> scout performing", "show me recent scout runs", "why did this scout find/report nothing", "what has the fleet learned", "explore scout run <id>", "is my scout working".
Use LLM as a simulator of expert debates and opposing viewpoints instead of getting a single sycophantic answer. Use this skill when the user needs to make an important decision, wants to stress-test an idea, needs devil''s advocate analysis, wants to avoid confirmation bias, or says "argue against this", "what are the counterarguments", "simulate experts", "debate this", "challenge my thinking". Based on Karpathy 31k-like post.
Use lasso, the web UI and MCP server over herdr, to spawn and manage coding agents across machines, notify the human who runs it, show them a file, and drive the shared browser they watch live. Use when you have lasso's MCP tools (create_agent, list_agents, notify, open_file, shared_browser, ...) or the `lasso` CLI, when asked to "spin up an agent", "start a worktree agent", "close yourself", "ping me", "open it for me", "show me in the browser", or when $HERDR_PANE_ID is set inside a lasso-created pane.
Create, validate, and consume Google Open Knowledge Format (OKF) v0.1 bundles — YAML-frontmatter Markdown knowledge atoms for portable, interoperable AI-agent sharing. Routes: use `okf-agent-memory` for the separate OKF v0.2 Go CLI/MCP project-memory runtime, `llm-wiki` for raw source capture and vault maintenance, `obsidian` for vault automation, `graphify` for durable committed graphs, and `scrapling` for web-content extraction. Triggers on: Google OKF v0.1, open knowledge format, knowledge bundle, OKF document, LLM wiki standard, knowledge atom, agent context format, OKF frontmatter, OKF bundle, knowledge interoperability
Deep retro of retros - investigation pass that surfaces improvements across older retros and the current ax setup. External AI agent drives the reasoning with high thinking enabled. Triggers when user says "deep retro", "investigate ax", "what should I fix in my setup", "retro of retros", or invokes /ax:retro-meta. Use after `ax:retro` if proposals are sparse or you want broader exploration.
Teaches agents to recognize and avoid security threats during normal activity. Covers phishing detection, credential protection, domain verification, and social engineering defense. Use when building or operating agents that access email, credential vaults, web browsers, or sensitive data.
Coordinate multiple terminal coding agents through the Superset CLI by creating isolated workspaces, launching workers, sending follow-ups, reading terminal output, tracking dependencies, and collecting structured results. Use when asked to delegate or parallelize coding work, coordinate agents across workspaces or hosts, hand work between agents, monitor worker progress, or run a multi-agent workflow with Superset. Do not use for ordinary single-agent workspace, task, or automation management.
Use when the user explicitly asks for vibe brainstorming, creative implementation ideas, implicit expected behavior, or convention checks, and when an implementation task is creative or convention-dependent. Do not use for obvious mechanical edits.
Batteries-included agent component for React/Next.js from ui.inference.sh. One component with runtime, tools, streaming, approvals, and widgets built in. Capabilities: drop-in agent, human-in-the-loop, client-side tools, form filling. Use for: building AI chat interfaces, agentic UIs, SaaS copilots, assistants. Triggers: agent component, agent ui, chat agent, shadcn agent, react agent, agentic ui, ai assistant ui, copilot ui, inference ui, human in the loop
Build automated AI workflows combining multiple models and services. Patterns: batch processing, scheduled tasks, event-driven pipelines, agent loops. Tools: inference.sh CLI, bash scripting, Python SDK, webhook integration. Use for: content automation, data processing, monitoring, scheduled generation. Triggers: ai automation, workflow automation, batch processing, ai pipeline, automated content, scheduled ai, ai cron, ai batch job, automated generation, ai workflow, content at scale, automation script, ai orchestration
Split a completed PostHog task run into activity records — what the agent tried, whether it worked, what blocked it — and record each one through the report_activity tool. Use when a task asks to analyze a run, produce a task analysis, or review a run from an attached run log. Covers the log query protocol (bounded jq queries over the raw JSONL), both log schemas, the activity schema, and evidence verification. Records facts only; it does not suggest fixes.
Connect the Vercel AI SDK to iMessage. Build AI agents and assistants that communicate over iMessage using chat-adapter-imessage, the official Photon adapter. Local mode (runs on your Mac) and remote mode (Photon's production infrastructure). Covers createiMessageAdapter, postMessage, editMessage, deleteMessage, react, startGatewayListener, types, and integration with the Chat SDK. Keywords: vercel ai sdk, imessage, chat adapter, ai agent, chatbot, conversational ai, messaging, apple messages, vercel, nextjs, ai assistant, chat sdk, real-time, macos.
Migrate prompts and code from Claude Sonnet 4.0, Sonnet 4.5, or Opus 4.1 to Opus 4.5. Use when the user wants to update their codebase, prompts, or API calls to use Opus 4.5. Handles model string updates and prompt adjustments for known Opus 4.5 behavioral differences. Does NOT migrate Haiku 4.5.
Rank an existing set of context chunks by relevance, diversity, freshness, and utility. Use when retrieval has already produced candidates that must be scored or reranked; use context-retrieval when the source corpus still needs to be searched.
Build a starting-point taxonomy for an MCP tool — what users were trying to accomplish before they reached the tool — and publish it as a PostHog notebook. Reconstructs each session's goal from its opening tool calls, then clusters those goals into named categories with size, share, and facet mix. Use when the user asks "why do people use this tool?", "what are users actually trying to do?", "what problem brings people here?", "where do these sessions start?", "segment usage of <tool> by goal", or wants a Clio-style taxonomy of MCP usage. Complements exploring-mcp-intent-clusters, which groups what agents did per call rather than why the session began. The agent running this skill writes the goal labels itself, reading the corpus query output session by session — the bundled scripts cover the mechanical facets but measurably lose the goal's altitude, so do not delegate that field to them.
Multi-engine long-term memory for game AI agents. Deep Godot 4.7+ (99 skills, 27 genre blueprints) + Bevy docs-first parity — audited patterns, persona orchestration (Analyst · Auditor · Builder), production-grade engineering.
Author-then-adversarial-review authoring pass for a single OpenSpec artifact — a subagent drafts it, `hrt-artifact-lint` clears the deterministic structure checks, then two subagents review it concurrently in fresh isolated contexts (a Destructive Critic for content/logic, a Structural Auditor for the structural judgement calls a linter cannot make), findings resolved before finalizing. Parameterized by artifact (specs or design). Use during the tempa-spec schema's `specs` and `design` phases.
Use when the context window fills too fast, per-turn input token cost is too high, or the aggregate input token bill needs explaining and fixing. Two audits: (1) overhead — tool schemas, skills index, system fixed text, user profile, memory recall measured live, ranked, reduced; (2) cost accounting — total input token volume across sessions, cron fires, and tool traffic, with cache-aware interpretation, pre-checked config levers, and scheduled delta checks. scripts/audit_overheads.py for per-turn overhead, scripts/audit.py for totals.
Read Cursor / Claude Code / Codex agent logs and call write_knowledge for each note found. Default auto-writes then reports what was cached and expected token savings (analytics-style: rediscovery/generation cost reused on each future read), and opens the HTML report. Dry-run lists the exact write_knowledge payloads (title, content, repo, branch) without writing. Use when the user says "Please bootstrap my knowledge with Dosu", "bootstrap agent knowledge", "/bootstrap-agent-knowledge", "log to dosu knowledge", "mine my sessions into Dosu", "backfill branch notes from my agent logs", "save my agent logs to Dosu", or wants a one-shot pass over local histories.
Rigorous engineering orchestrator ported from Lauren Tan's pstack (poteto-mode): reads your task, picks one of 23 playbooks (bug fix, feature, refactoring, perf, investigation, prototype, babysit, shipping, autonomous run, orchestrate, and more), routes to bundled procedures (how, why, architect, arena, swarm, interrogate, unslop, technical-writing, show-me-your-work, tdd, and others), and applies 24 engineering principles. Self-contained: no plugin install, no sibling skills required, works with any agent that reads skills.sh-format SKILL.md files. Use whenever a task needs rigor: nontrivial code changes, architecture decisions, debugging, reviews, PRs, long autonomous runs, or "work like poteto", "poteto-mode", "pstack".
Orchestrate the full Roblox game development lifecycle. Use as the umbrella entrypoint for broad requests like create a Roblox game, build a new game, continue development, what should we do next, take this from idea to launch, make this game profitable, plan the roadmap, manage production, or maintain a live Roblox game. Routes to the right Roblox specialist skills by GDLC phase.
Sequential reasoning with deep self-reflection and backtracking. Use when problems have step-by-step dependencies, need careful logical reasoning, or require error correction. Each step includes self-reflection, and incorrect steps trigger backtracking. Ideal for debugging, mathematical proofs, sequential planning, or causal analysis where order matters.
Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy
Execute approved, bounded work in an isolated implementation conversation and return a structured evidence receipt. Use with inherited history or an explicitly identified source; no particular planning, testing, or review skill is required.
Create, validate, and transition documentation artifacts (Vision, Initiative, Epic, Spec, Spike, ADR, Persona, Runbook, Design, Journey) through lifecycle phases. Handles spec writing, feature planning, epic creation, initiative creation, ADR drafting, research spikes, persona definition, runbook creation, design capture, architecture docs, phase transitions, implementation planning, cross-reference validation, and audits. Also invoke to update frontmatter fields, re-parent an artifact under a different epic or initiative, or set priority on a Vision or Initiative. Chains into swain-do for implementation tracking on SPEC; decomposes EPIC/VISION/INITIATIVE/JOURNEY into children first.
Update swain skills to the latest version. Use when the user says 'update swain', 'upgrade swain', 'pull latest swain', 'reinstall swain', 'refresh skills', or wants to update their swain skills installation. Uses npx to pull the latest swain release from GitHub, with a git-clone fallback, then invokes swain-doctor to reconcile governance and validate project health.
Apply the Theory of Constraints (Goldratt's Five Focusing Steps) to find and fix the single bottleneck that caps a process, workflow, pipeline, or Agent Skill/plugin's throughput. Use whenever optimizing speed, lead time, cost, or token/context budget of a system — CI/build pipelines, dev value streams (idea→review→merge→ship), an Agent Skill's trigger/context flow, or runtime code paths — ESPECIALLY when you don't know where to optimize, a local speedup didn't improve the whole, work piles up at a stage, everything is busy but little ships, adding capacity didn't help, or a policy/rule (not a resource) is the limiter. It locates the constraint with measurement, then prescribes exploit → subordinate → elevate → repeat, and stops you optimizing non-constraints (the "mirage of the non-bottleneck"). Trigger even when the user just says "speed this up", "why is this slow", or "make this more efficient" without mentioning constraints or bottlenecks.
Work with PostHog Signals scouts: scheduled agents that monitor a project and write reports into the Signals inbox. Use to assign monitoring work, schedule quality scoring, find which scout covers a surface, act on reports, reduce noise, investigate missing findings, or improve the fleet through feedback and calibration. Also covers follow-up checks that verify whether a reported problem stays fixed. Use `authoring-scouts` for edits, `exploring-scouts` for run observability, and `inbox-exploration` for report triage. Trigger on "work with my scouts", "get more out of scouts", "have a scout watch X", "tell me if Y spikes", "score X on a schedule", "measure quality of Y", "what do I do with this scout report", "calibrate/review my scout fleet".
Browser automation CLI for AI agents using Vercel's agent-browser. The best tool for AI-driven browser automation — uses deterministic refs from accessibility trees instead of fragile selectors. Optimized for LLMs with fast Rust CLI, JSON output, and purpose-built AI workflows. Use when you need reliable, scriptable browser automation that just works.
Route broad or ambiguous AgentKit SEO work to the right module while keeping context scoped. Use when a request spans multiple surfaces, asks for overall digital-presence strategy, involves provider or install architecture, needs agent-context planning, or the correct platform skill is unclear.
Use when running, configuring, or troubleshooting the autonomous-dev-team dispatcher cron. Triggers on phrases like "run the dispatcher", "scan for pending issues", "dispatch autonomous tasks", "set up the dispatch cron", "configure dispatcher.conf", "set up multi-project dispatcher", "dispatch to a remote dev box via SSM", "EXECUTION_BACKEND=remote-aws-ssm", "stale agent detection", or working on dispatcher-tick.sh / dispatcher-multi-tick.sh / dispatch-local.sh / dispatch-remote-aws-ssm.sh. Covers per-project tick (5 steps: concurrency, scan-new, scan-pending-review, scan-pending-dev, stale detection), the multi-project outer loop, and pluggable local-vs-remote-AWS-SSM execution backends.
Generate detailed implementation plans for complex tasks. Creates comprehensive strategic plans in Markdown format with objectives, step-by-step implementation tasks using checkbox format, verification criteria, risk assessments, and alternative approaches. All plans MUST be validated using the included validation script. Use when users need thorough analysis and structured planning before implementation, when breaking down complex features into actionable steps, or when they explicitly ask for a plan, roadmap, or strategy. Strictly planning-focused with no code modifications.
Orchestrate approved, bounded work through a same-directory Codex App fork: create and name the execution task, ask it to run spec-executor, route decisions through Codex Task Messenger, validate the returned receipt, and archive a completed child. The source may come from any planning workflow. Requires Codex App task tools and codex-task-messenger.
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.