Skills
97,501 packaged capabilities and playbooks for agents. Install any of them with one command.
Agent workflows
Reviews Deep Agents code for bugs, anti-patterns, and improvements. Use when reviewing code that uses create_deep_agent, backends, subagents, middleware, or human-in-the-loop patterns. Catches common configuration and usage mistakes.
Selects sufficient, low-noise local shell commands for repository exploration, git inspection, tests, and container checks. Use when choosing command scope or output detail, especially for token savings or terse workflows; not for incidental tool mentions or requests for full output.
Build practice environments and evaluation gyms where agents can try, fail, and learn from feedback. Use this skill when the user wants to set up automated evaluation for an agent, needs a sandbox for testing LLM capabilities, wants to build a practice gym for skills, needs an RL-style environment for agent improvement, or says "eval environment", "practice gym", "agent sandbox", "automated testing for agents", "RL environment", "agent eval loop". Based on Karpathy LLM Textbook and environments posts.
Take your AI agent to the next level with full LangWatch integration. Adds tracing, prompt versioning, evaluation experiments, and simulation tests in one go. Use when the user wants comprehensive observability, testing, and prompt management for their agent.
Blind side-by-side multi-model comparison. Send one prompt to 2-4 models simultaneously, present responses anonymously (Model A / B / C / D), let the user pick a winner, then reveal identities and show which model won. Supports custom evaluation criteria, synthesis of responses, and vote history logging. Trigger when the user says "compare models", "test these models", "which model is better for", "A/B test", "blind comparison", "model evaluation", or wants to see how different AI models handle the same prompt. Can also be used for prompt engineering — testing how different models interpret the same instructions. Supports reasoning-effort A/B via provider:@effort:model_id specs.
(NS) Ad-hoc coding — bug fixes, refactors, scripts, migrations — without full SDD. Entry priority 5: \"just implement this\", \"quick fix\", or coding without execution-handoff (C2 under ns-autonomous). Do NOT: ISSUE_URL (ns-execution-gitlab-issue), multi-day/version (ns-spec-driven), product bug report / screen expected-vs-actual **without** diagnosis_complete (ns-bug-report-diagnosis), diagnosis-only stack/CI (ns-investigator), when execution-handoff.md exists. Do NOT generate requirements/tasks/handoff.
Audit and reorganize a Claude Code / Codex skill library — enforce a naming convention (prefix each skill by which backend or product it touches), rename skills consistently, fix the symlinks and cross-references a rename breaks, and report broken symlinks and stale name references. Use when the user says "organize my skills", "rename these skills", "clean up the skill library", "enforce a skill naming convention", or "fix broken skill symlinks".
Front door of the SDD flow. When the user asks for a feature or bug fix ("implementa…", "agrega…", "arregla…", "add…", "build…", "fix…"), FIRST scan the request + codebase context for load-bearing ambiguity and ask a few targeted clarifying questions (selectable options, recommended first) BEFORE writing any spec or code — then hand a well-formed goal to /opsx:propose (small) or the sdd-feature-flow workflow (large). Skip the questions when the request is already unambiguous. Modeled on GitHub Spec-Kit's /clarify.
Author or update the project's constitution — openspec/project.md — the non-negotiable principles, conventions, and constraints the Spec-Driven Development flow must honor. Modeled on GitHub Spec-Kit's /constitution. Run once per repo when onboarding SDD, or whenever the team's principles change. The sdd-feature-flow Analyze gate checks every spec against it.
Ingest raw context the user pastes or points at — a ticket, a design doc, meeting notes, a spec, a URL, referenced files/paths — and have an agent READ and UNDERSTAND all of it, then synthesize a well-formed feature brief (goal, scope, constraints, and load-bearing unknowns) that feeds sdd-clarify and the sdd-feature-flow harness. Use at the very start of a feature when you have source material instead of a one-line goal, or whenever the user says "here's the context" / "read this" / dumps a ticket or doc.
Install + verify ax (the agent experience layer). Triggers when the user says "install ax", "set up ax", "ax not found", "ax doctor", "is ax running", "fix ax install", "first-time ax setup", or any setup question about the ax CLI / skills / daemon. Walks the install via the install.sh + skills.sh + first ingest, validates with `ax doctor`, and points the user at ax:retro (experiment loop) and ax:extract-workflow (reconstruct workflow behind shipped artifacts).
Create, update, validate, or reorganize repo-local Agent Skills under .agents/skills, including SKILL.md frontmatter, first-level references, compatibility symlinks, and validation scripts. Use only for skill-tree maintenance, skill taxonomy changes, or agent-guidance validation work.
Use when the user says "brief this task", "scope this properly", "what do you need from me", "improve my prompt", "help me get better output", or hands over a SUBSTANTIAL task — multi-step, multi-document, or a deliverable someone else will read — as a thin one-line request. Compile a task brief BEFORE starting substantial work — pin down the goal (what "done" looks like), context, constraints (standing rules included automatically), and tooling — confirm with the user, then execute against it. Asks at most 2-3 clarifying questions, only where the answer would change the output; otherwise states assumptions in the brief and proceeds. Not for quick one-step asks — answer those directly.
Venice augmentation endpoints for agent pipelines. Covers POST /augment/text-parser (extract text from PDF/EPUB/DOCX/PPTX/XLSX/XLS, plain text and source code; multipart, up to 25MB; JSON or plain-text response), POST /augment/scrape (fetch a URL and return markdown; blocks X/Reddit and private/internal URLs), and POST /augment/search (Brave ZDR or anonymized Google; structured title/url/content/date results, limit up to 20 but effectively capped at 10 Brave / 5 Google). $0.01 per successful call, 20 req/min on scrape and search, privacy, and error shapes.
Expert in designing, orchestrating, and managing multi-agent systems (MAS). Specializes in agent collaboration patterns, hierarchical structures, and swarm intelligence. Use when building agent teams, designing agent communication, or orchestrating autonomous workflows.
Scaffold a custom MCP (Model Context Protocol) server that integrates with Autodesk Platform Services (APS). Use this skill when the user wants to build, create, generate, or set up an MCP server for APS, Autodesk, Forma, or any Autodesk cloud API. Supports Node.js/TypeScript, .NET/C#, and Python. Handles STDIO (local) and Streamable HTTP (cloud) transport, and various APS auth patterns: 2-legged OAuth, Secure Service Accounts, and 3-legged OAuth.
Solutioning flagship — shards a PRD + architecture into epics.md and individual {epic}.{story}.{slug}.story.md context objects, the LAST planning artifact before external dev handoff. Each story is a self-contained ~8K-token compiled context object: Dev Notes with SOURCE CITATIONS back to prd.md/architecture.md, Acceptance Criteria, Tasks/Subtasks mapped to ACs, Testing strategy, Dependency Maps, an explicit Owned File/Module Scope list (the lever for parallel-conflict-free scheduling), and Learnings from Previous Stories. Sized to one dev-day; split if larger; NO story points. Use when the user says "shard the PRD", "create epics", "break the PRD into epics", "break this epic into stories", "create stories", "create a story", "draft story files", "generate the story for X", "prepare stories for dev", "mark the story ready for dev", "validate this story", or "what stories are in this epic". Three intents: Create new epics/stories, Update an existing story, or Validate a draft against the contract.
Facilitate structured idea exploration and product/design specification. Use when a user wants to talk through an idea, refine it via iterative questions, and converge on a clear design/spec (and later an implementation plan), especially after inspecting the current project state.
Run chief-agent in full autopilot. Requires goals and contracts to exist. Chief creates TODO, delegates to builder, and repeats until milestone is done. Auto mode makes all decisions autonomously; safe mode stops on ambiguity. Use "/chief-autopilot" for auto or "/chief-autopilot safe" for safe mode.
Orchestrate the full workflow for analyzing plot points and dramatic functions in long texts: preprocessing, segmented analysis, and result integration. Use when a novel or script needs a structured dramatic analysis report, or when a dramatic workflow or plot-point analysis pipeline is mentioned.
Read the graph BEFORE any answer, advice, retelling of someone else’s advice, or claim that “it isn’t in the graph” about something that may be recorded: read the node’s neighborhood and follow the relevant links; answer from the graph, not training data or code. Triggers: “get oriented,” “enter the graph,” “what does the graph say,” “in the graph,” “we discussed,” “from my notes,” NKS, Искрон, вимарша, крия, контур, роль, феномен, бяньхуа, эстафета, an unfamiliar term from the user’s vocabulary; “what is kriya,” “explain the methodology.” Use before answering anything about the user’s Iskron graph
This skill should be used when the user asks to "execute this blueprint", "execute the blueprint", "run the blueprint", "run the plan", "execute the plan", "start building from the plan", "implement the blueprint", "implement the plan", "continue the plan", "resume execution", "execute 01_milestone_name.md", "execute step 3", "run steps 3-5", or "skip to step 4".
Improve an existing skill in the IC Skills repo (skills.internetcomputer.org). Load this skill whenever asked to improve, fix, update, enhance, review, or add content to an existing skill at skills/<skill-name>/SKILL.md — including adding pitfalls, updating canister IDs, fixing code examples, strengthening the description, or updating evals. Do NOT use for creating a brand-new skill (use skill-creator for that). Trigger examples: "the motoko skill is missing a pitfall", "update icp-cli for the new recipe format", "the ckbtc description is weak", "add an eval for the canister-security skill".
Read this BEFORE launching any subagent (Task tool, background agents, parallel agents, best-of-N, delegating work to another agent). Consensus principles for using subagents well. Triggers: launch a subagent, spawn agents, run agents in parallel, delegate to a subagent.
Zero-config goal-to-tasks engine (the Atlas engine). Takes any goal (software, pentest, business, learning), runs adaptive discovery via brainstorming, generates a validated spec, parses into TaskMaster tasks, and hands off to execution. Use when user says "PRD", "product requirements", "I want to build", invokes /atlas, or wants task-driven development.
A comprehensive auditor for any agent skill — including Manus, OpenClaw/ClawHub, Claude, LobeHub, or custom SKILL.md-based skills. Use this skill whenever a user wants to evaluate, audit, review, score, or quality-check an agent skill before publishing, updating, or deploying. Covers two hard veto gates (structural redlines + research integrity redlines), static quality scoring across 25 criteria (ISO 25010 + OpenSSF + Agent), dynamic test input generation, multi-mode execution testing, multi-layer output evaluation with five specialized category rubrics (Evidence Insight / Protocol Design / Data Analysis / Academic Writing / Other), a Research Veto that applies to all four research categories, human eval viewer generation, actionable P0/P1/P2 optimization recommendations, and automatic skill improvement that outputs a polished, production-ready SKILL.md. Also use whenever a user says "audit my skill", "evaluate my skill", "improve my skill", or wants a corrected version after evaluation.
Create and edit agent skills with enforced conventions. Use when creating a new skill, modifying an existing skill, or harmonizing skill structure. Also use when the user mentions "skill conventions", "clean up skills", "standardize skills", or wants to ensure skills follow best practices for token efficiency and triggering accuracy.
Connect to the Superset MCP server to create Git-worktree workspaces, launch coding-agent sessions, schedule automations, and manage tasks on a user's Superset account. Use when an agent needs to orchestrate parallel coding agents programmatically over MCP instead of the CLI.
Task tracking and implementation execution for swain projects. Invoke whenever a SPEC needs an implementation plan, the user asks what to work on next, wants to check or update task status, claim or close tasks, manage dependencies, abandon work, bookmark context, or record a decision. Also invoked by swain-design after creating a SPEC that's ready for implementation. Tracks SPECs and SPIKEs — not EPICs, VISIONs, or JOURNEYs directly (those get decomposed into SPECs first). Triggers also on: 'bookmark', 'remember where I am', 'record decision'.
agent-deck, the terminal session manager for AI coding agents. Use when the user mentions "agent-deck", "session", "sub-agent", "MCP attach", or "git worktree", or needs to create, start, message, read, fork, restart, or supervise agent-deck sessions; launch child sessions or consult another CLI through agent-deck; attach MCPs to a session; manage groups, profiles, accounts, or worktree sessions; set up conductors or watchers; configure config.toml; or troubleshoot agent-deck. Not for in-conversation subagents, plain git worktrees, or plain tmux.
Audits all OrchestKit skills for quality, completeness, and compliance with authoring standards. Use when checking skill health, before releases, or after bulk skill edits to surface SKILL.md files that are too long, have missing frontmatter, lack rules/references, or are unregistered in manifests.
Reconstruct the workflow behind a past artifact - "what made X work" / "extract workflow from <date|sha>" / "how did we ship Y". Uses ax to find the relevant sessions and narrate the ordered skill arcs that produced the result. Triggers on "what made X work", "how did we build Y", "extract workflow from <date>", "extract workflow from <sha>", "what was the workflow around <topic>", "show me how I shipped <feature>", "reconstruct the recipe for <artifact>". Do NOT fire on generic "show recent activity" - this skill is for reconstructing the workflow behind a specific artifact.
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.