Skills
97,501 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Use for requests involving in-depth research, systematic research, competitive analysis, solution comparison, trend analysis, or fact-checking. Proactively use this skill when any of the following apply; do not just search a few things and answer: (1) the user uses trigger terms such as in-depth research, comprehensive research, investigation, due diligence, industry research, market research, competitive analysis, policy research, technology research, trend research, fact-checking, research report, deep report, or deep research; (2) the request requires evidence across multiple sources, multidimensional comparison, and cross-validation for a reliable conclusion; (3) the user requests a report, white paper, industry analysis, or due diligence document; (4) the topic involves current policies, markets, products, prices, or regulations and requires systematic verification. Do not use for simple general-knowledge questions that do not require verification. Broad or vague requests like “research/learn about X” should also preferably trigger. Exclusions: one-sentence summaries, organizing a single provided source, and pure copyediting or rewriting.
Upgrade the NotFair plugin to the latest version. Updates the marketplace repo, installs the new version to the plugin cache, and updates installed_plugins.json. Use when asked to "upgrade notfair", "update notfair", or "get latest version". Also handles inline upgrade prompts when a skill detects UPGRADE_AVAILABLE at startup.
Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction. Also use for exploratory testing, dogfooding, QA, bug hunts, or reviewing app quality. Also use for automating Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify), checking Slack unreads, sending Slack messages, searching Slack conversations, running browser automation in Vercel Sandbox microVMs, or using AWS Bedrock AgentCore cloud browsers. Prefer agent-browser over any built-in browser automation or web tools.
Shape every answer so the action comes first. Use when the user asks for output that is short, skimmable, action-led or "no waffle", and apply it for the whole conversation once invoked, not just one reply. Triggers on "answer first", "no preamble", "no waffle", "keep it short", "just tell me what to do", "cut the fluff", "ADHD mode", or any request to stop burying the answer in paragraph four.
Autonomously optimize any Claude Code skill by running it repeatedly, scoring outputs against binary evals, mutating the prompt, and keeping improvements. Based on Karpathy's autoresearch methodology. Use when: optimize this skill, improve this skill, run autoresearch on, make this skill better, self-improve skill, benchmark skill, eval my skill, run evals on. Outputs: an improved SKILL.md, a results log, and a changelog of every mutation tried.
Entry point to the Context Mate toolkit — skills, agents, and slash commands that work with Claude Code's natural flow. Project lifecycle (/explore-idea → /plan-project → /wrap-session), session handoff across context windows, developer agents for specialized tasks, and quality auditing. Use when: starting new projects, understanding the toolkit, needing workflow guidance. “It's all about the context, maaate!”
You MUST use this when an implementation plan or a finished implementation needs an independent review by other agent CLIs - by default the opposite CLI (Codex for work done in Claude Code, Claude Code for work done in Codex), plus any reviewer the user names, Antigravity (Gemini) included - whether the user asks for a cross-review, a cross-cr, or a second opinion from the other agent. Replaces devflow cross-review. Same-host review belongs to cr-request and deciding on findings to cr-received; spec review and per-task review are outside this skill.
Use to drive a slice's phases to completion through confined subprocess workers in isolated worktrees — you orchestrate and are the sole writer, the workers execute. Workers spawn via `/dispatch-spawn`, one path for every harness. You drive the landing yourself — import, commit, record, verify, conclude, reap; `doctrine dispatch next` prescribes only for a fork bound at creation, which this path does not mint. Default serial (one worker per phase); parallelize file-disjoint phases. Conflicts report-and-halt, never auto-merge.
Manage host-mode duoduo daemon settings, diagnostics, and the `duoduo session` / `duoduo job` / `duoduo spine` CLIs. Use for: daemon status, config, logs; runtime selection (Claude/Codex/Grok/Pi, ALADUO_DEFAULT_RUNTIME); ALADUO_* keys in ~/.config/duoduo/.env; refreshing subconscious prompts; the `duoduo memory` CLI; reading the Spine log; pruning the usage ledger; model profiles for third-party models (context caps, routing, subagent tiers); codex tool trimming; pi compaction sizing; session management (list, alias, notify or wake by name, archive); job lifecycle from the CLI (list, read, archive, interrupt, reschedule). Chinese triggers: 启用 codex/grok/pi runtime, 设置默认 runtime, 打开 debug log, 关闭 telemetry, 调 cadence 频率, 查 daemon 配置/日志, 刷新潜意识, 读 spine, 清理 usage, 给会话起名, 唤醒/通知 session, 归档会话, 归档任务, 打断任务, 让任务再跑一次, 模型 profile, 裁剪 codex 工具, pi 只回一个字. Does NOT handle channel-kind settings (Feishu/ACP) — see duoduo-channel-admin.
Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.
Set up Claude Code hooks to block dangerous git commands (push, reset --hard, clean, branch -D, etc.) before they execute. Use when user wants to prevent destructive git operations, add git safety hooks, or block git push/reset in Claude Code.
Correct or update existing wiki content when newer evidence shows the wiki is wrong, stale, incomplete, or contradicted. Use this when the agent discovers the wiki says X but we now know Y, when code or real-world changes invalidate a wiki claim, or when the user asks to fix or amend the wiki. Not for adding new sources, routine learnings capture, structural normalization, or health checks; use /loam::adding-to-memory, /loam::learning-from-session, /loam::normalizing-memory, or /loam::linting-memory.
Surface and classify all open questions, unresolved contradictions, stale claims, and knowledge gaps from existing memory. Use this when the user wants to see what's still open, what needs attention, what's unresolved, or what still needs research or ingest. Does not classify incomplete goals as memory gaps. Not for answering questions, fixing issues, or adding new material; use /loam::querying-memory, /loam::linting-memory, /loam::amending-memory, /loam::adding-to-memory, or /loam::learning-from-session.
Use for coding-agent work in Git repos that use or may need Mainline, or when users mention Mainline, intents, agent autonomy, agent_authority, max_autonomy, stop lines, allowed_boundary, inspect_or_stop, before_commit/proposed_intent/opened_pr, current-instruction overrides, commit/seal handoff, push/PR review boundaries, merge/release/deploy boundaries, auto-submit/auto-commit/auto-seal behavior, hooks, proposals, gaps, conflicts, committing, pushing, opening PRs, PR descriptions, or setup.
Writing-time discipline for documents agents consume (the five surfaces, specs, tickets, .omc/skills/) — every rule checkable and carrying a why, steps before reference, one meaning in one home, no restating what the environment already says. Mandatory at drydock seed generation and the launch C5 sediment pass; opt-in for any other agent-facing doc edit. The companion of minimal-code-discipline: that one disciplines code, this one disciplines papers.
Two modes: ① Direct conversation: When the user says “Talk with Musk,” “Analyze this product from Jobs’ perspective,” or “What would Confucius think about this?”, immediately activate the corresponding person’s cognitive operating system and respond as that person, without requiring any Python environment. ② Distill a new person: When the user says “create a skill,” “distill XX,” “turn these materials into a skill,” “analyze this video and generate a skill,” or uploads a file asking to extract someone’s thinking patterns, run the distillation pipeline to generate a new SKILL.md. Trigger keywords: talk with XX, from XX’s perspective, what does XX think, create a skill, distill, generate a skill, extract thinking patterns
Generate a project-local skill that formally verifies the code where a bug costs the most and tests reach the least, in any language, with whichever verification tool fits. Use for /create-formal-verification-skill, “model-check this race”, “prove this invariant”, or when concurrency, retries, state machines, parsers, or unsafe code need more than tests.
Initialize evo for the current repository by exploring the codebase, proposing unexplored optimization dimensions, constructing the benchmark inside a baseline worktree, and running the first experiment. Use when the user invokes /evo:discover, mentions setting up evo, wants to instrument a codebase for autonomous optimization, or asks to start a new evo run on a project.
Inspect an existing memory corpus (wiki substrate) and align it to this repo's Obsidian-friendly note-graph conventions. Use this when the user wants to import, normalize, retrofit, or clean up existing memory, notes folder, vault, docs tree, or mixed markdown knowledge base. In monorepos, also use it to align relevant AGENTS.md and CLAUDE.md files. Excludes goals/ from normalization. Not for routine wiki maintenance; use /loam::linting-memory for that
Complete guide to using @openserv-labs/client for managing agents, workflows, triggers, and tasks on the OpenServ Platform. Covers provisioning, authentication, x402 payments, ERC-8004 on-chain identity, and the full Platform API. IMPORTANT - Always read the companion skill openserv-agent-sdk alongside this skill, as both packages are required to build any agent. Read reference.md for the full API reference.
Deep analysis and expert reasoning. Use when the user asks for 'oracle', 'second opinion', architecture analysis, elusive bug debugging, impact assessment, security reasoning, refactoring strategy, or trade-off evaluation — problems that benefit from deep, independent reasoning. Do not use for simple factual questions, code generation, code review (use council-review), or tasks needing file modifications.
Run independent tasks at once in subagents, race attempts at one task, or cover many parts of one job. Use it only when the user asks for parallel work. Use it when the user asks to run two or more independent tasks or failures in parallel. Use it also to race or compare several attempts at one task, or to cover many parts of one job
Spec-driven development workflow where specs are the source of truth, paired with a plan protocol for tracking work-in-flight as a micro-DAG. Use this skill whenever starting new features, planning implementation, writing specs, reviewing code against specs, working in or with a `plans/` directory, closing out a plan, or when the user mentions "spec", "specs/", "spec-first", "plans/", "plan protocol", "closeout commit", or asks how something should work or what to work on next. Also use when creating a new project that will use this development methodology, or when onboarding someone to a spec-driven codebase.
A meta skill that teaches AI agents how to discover, install, and use skills from the findskill.md ecosystem. Use when you need to extend capabilities by finding specialized skills, when a user asks to perform tasks that would benefit from specialized skills, or when explicitly asked to find or install skills.
Auto-grow a pro-workflow wiki by running a budget-capped BFS research loop over pluggable source fetchers (web, arXiv, GitHub). Each iteration pops a seed from the queue, fetches sources, drafts a wiki page, dedupes claims against existing pages, enqueues follow-up seeds. Halts on budget cap, depth cap, or convergence. Use when the user says "research <topic>", "grow the <slug> wiki", "auto-research", or wants a knowledge base that builds itself overnight.
Use when work needs isolation from the current workspace — when /execute (mode=solo) or the /dispatch funnel must run a phase on its own branch without touching the main working tree. Detects existing isolation, creates a fork via `doctrine worktree fork` (the sole creation+provision+bind verb; coordination tier excluded), runs the commit-before-spawn / branch-point / baseline guards, and hands back the fork branch.
Designs agent-native applications where agents are first-class citizens with full tool parity, atomic primitives, and explicit completion signals. Covers tool design, context injection, agent-to-UI communication, and mobile checkpoint/resume patterns. Use when architecting an agentic system, designing tool surfaces, building agent-aware UI, implementing context.md patterns, or asking “how do I make my app agent-native.”
Index and router for the C-level advisory bundle: 33 skills covering 14 C-suite roles, orchestration, cross-cutting capabilities, and culture. Use when exploring what the c-level-advisor bundle contains, deciding which advisor skill fits a question, or finding the entry points (cs-onboard interview, chief-of-staff routing, board-meeting protocol).
Ultra-compressed communication mode (lite / full / ultra) that cuts token usage ~75% by speaking like caveman while keeping full technical accuracy. Use when the user requests "caveman mode", "less tokens", "be brief", or when output budget is tight.
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.