Evaluate a FlowGoal against its evidence ledger and update lifecycle status to one of {pass, incomplete, fail, needs_human_review, blocked} by running deterministic verification commands first, then (when stopHookEnforcement=evaluator-loop or explicit /flow:goal evaluate invocation) dispatching the goal-evaluator-judge agent for fuzzy rubric criteria. Use when /flow:goal evaluate is invoked, when the Stop hook fires in evaluator-loop mode, or when /flow:start Phase 4 needs to convert AC evidence into a verdict. This skill MUST be consulted because lifecycle transitions without deterministic evidence enable silent premature completion — the goal contract is only as good as the evaluator that proves or disproves it.

Use when presenting learning recommendations from background analysis, reviewing pending pattern suggestions, or approving automation scripts from traces. Load when learning-agent has findings to present.

Add OAuth 2.0 PKCE authentication to a remote MCP server. Use this skill whenever the user wants to add authentication to an MCP server, protect MCP tools with OAuth, implement login flow for an MCP connector, add user auth to an MCP endpoint, or set up token-based access for MCP. Also triggers on: 'MCP OAuth', 'MCP authentication', 'withMcpAuth', 'MCP login flow', 'protect MCP endpoint', 'MCP token auth', 'dynamic client registration MCP', 'Claude connector OAuth'. Even if the user just says 'add auth to my MCP server' or 'my MCP server needs login', use this skill.

Knowledge accumulation and retrieval patterns for file-based agent memory

Comprehensive guide to Output.ai Framework for building durable, LLM-powered workflows orchestrated by Temporal. Covers project structure, workflow patterns, steps, LLM integration, HTTP clients, CLI commands, and the full inventory of available agents, commands, and skills.

Domain-neutral methodology for autonomously extracting structured notes from a single academic paper via three escalating passes. Pass 1 (inspectional, ~10-15 min) reads title, abstract, intro, section headings, conclusion, references at a glance, then applies the Five Cs framework (Category, Context, Correctness, Contributions, Clarity). Pass 2 (content grasp, ~30-60 min) reads the full paper skipping proofs, answers main-argument / Big-Question / hypotheses / figure-by-figure / references / confusions. Pass 3 (deep understanding, ~1-4 hours, reserved for important papers) virtually re-implements, challenges every assumption, identifies what is NOT said, asks the falsifiability question. The methodology is internal to the agent applying it - questions are answered against the paper's content, never asked of the operator. Inspired by Keshav 2007 ("How to Read a Paper") and Adler-style inspectional reading. Use when an extraction agent needs to convert dense academic prose into structured machine-and-human-readable notes - bio papers, CS papers, ML papers, statistics, math, any field.

Internal sub-skill. Do not auto-activate. Use only when explicitly invoked by name by another skill or agent.

Start a new project from a request — explores, interviews, plans, creates milestones, and implements with agent swarms. Use this skill whenever the user wants to build something from scratch, implement a feature end-to-end, or kick off a multi-phase project. Triggers on phrases like "build me...", "implement...", "create a project for...", "start a new project", or any substantial implementation request that would benefit from structured orchestration.

Use this skill when the user says "project manager", "pm", "where are we", "what's in progress", "what should I work on next", "project status", "open issues", "what's next", "track this work", "review this issue", "improve this issue", "what are our conventions", "create a project", "plan this project", "project health", or asks about Linear issues, GitHub issues, GitHub PRs, or any issue tracking for the current project. Also use proactively when the user describes work that doesn't appear to be tracked, when starting new work on main/master, when branch naming suggests untracked work, or when the user has a spec they're about to file as a Linear initiative or GitHub project. Routes to focused pm-* skills (status, issues, templates, review, improve, conventions, branches, report, spec, handoff, setup, projects) based on the task. For autonomous multi-step work, delegate to the `pm-agent` sub-agent.

Search across company knowledge bases (Confluence, Jira, internal docs) to find and explain internal concepts, processes, and technical details. When an agent needs to: (1) Find or search for information about systems, terminology, processes, deployment, authentication, infrastructure, architecture, or technical concepts, (2) Search internal documentation, knowledge base, company docs, or our docs, (3) Explain what something is, how it works, or look up information, or (4) Synthesize information from multiple sources. Searches in parallel and provides cited answers.

Intelligently routes user requests to the most appropriate Claude Code skill. ALWAYS use this skill FIRST when user asks for help, mentions "skill", "which", "how to", or seems unsure about which approach to take. This is the default entry point for all skill-related requests.

Transform clichéd story elements by pushing along the emotional vector toward statistical edges. Use when first instincts are too predictable, when elements feel generic, or when you need the core methodology for avoiding statistical-center defaults.

Orchestration rules for strategy analysis workflows. Loaded when coordinating strategy agents, managing analysis phases, or running /strategy-report. Defines flows, stop points, agent coordination, and report compilation.

Mechanical documentation hygiene. Spawns a doc-maintainer agent to audit all project documentation and fix correctness, completeness, freshness, and consistency issues autonomously within its authority. Surfaces anything requiring user judgment for approval.

# Twilio SMS Skill - Text Message Bridge for Claude Code

Zoom Virtual Agent iOS integration via WKWebView. Use for Swift/Objective-C script injection, message handlers, support_handoff relay, and URL routing policies.

Create a ZIA URL Filtering rule that controls user access to web content by URL category, protocol, request method, user agent, user/group/department, location, device trust level, and optional time-of-day schedule (Time Interval). Supported actions: ALLOW, BLOCK, CAUTION, ISOLATE. Use when an admin asks to 'create a URL filtering rule', 'block category X', 'allow category Y', 'show a caution page for Z', 'isolate access to risky sites', 'block social media during work hours', or 'add a URL filtering exception'. Supports both recurring schedules (`time_windows`) and one-shot date-range validity (`enforce_time_validity`). This skill creates exactly one URL Filtering rule and chains to `zia-manage-time-interval` when the admin's request includes a recurring schedule.

Validate and register new Claude Code Skill manifests (.skill.yaml) to ensure structure, inputs/outputs, and dependencies are correct.

Adds one or more investor records to the Carta CRM via the Carta CRM MCP Server. Use this skill when the user says things like "add an investor", "/add-investor", "add investor to Carta CRM", "create investor record", "add this VC fund to the CRM", or "save investor data". Collects investor information conversationally, then creates it via the MCP server.

Adds a comment/note to a deal record in the Carta CRM via the Carta CRM MCP Server. Use this skill when the user says things like "add a note", "create a note", "log a note", "add note to a deal", "add note to CRM", "add note to Carta CRM", "log a comment on a deal", or "/add-note". Notes are stored as comments on deal records.

Use when deciding whether to launch an agent, selecting which agent to use, or coordinating multiple agents. Covers delegation criteria, external-model patterns, task isolation, and agent selection strategies.

Refactor bloated AGENTS.md, CLAUDE.md, or similar agent instruction files to follow progressive disclosure principles. Splits monolithic files into organized, linked documentation.

Refactor bloated AGENTS.md, CLAUDE.md, or similar agent instruction files to follow progressive disclosure principles. Splits monolithic files into organized, linked documentation.

Latest AI models reference - Claude, OpenAI, Gemini, Eleven Labs, Replicate

RAG-powered chat interface — retrieves relevant chunks via vector search, streams answers with source citations (page numbers + document titles), and persists conversations. Use this skill when the user says "add RAG chat", "chat with PDFs", "setup ai-rag-chat", or "add document Q&A".

Use this skill when a working tree contains uncommitted changes that need to be split into a sequence of atomic commits — typically after a long session, an agent hand-off, a rebase resolution, or anytime `git status` shows mixed work that was not committed as it landed. The skill assumes the agent has fresh context and did not author most of the diff, so it treats the working tree as evidence to investigate before grouping. Optimizes for `git bisect`: each commit is the smallest buildable and deployable unit, and no smaller. Uses `cortex git commit` for file-level commits and `cortex git patch --diff` when unrelated changes share a file.

AI citation readiness audit ONLY (does not touch Google rankings, use blog-rewrite for combined Google+AI work). Use whenever the user wants their content to rank in ChatGPT, Perplexity, Claude, Gemini, or Google AI Overviews. AI citation optimization audit scoring blog posts for ChatGPT, Perplexity, and Google AI Overview citability. Evaluates passage-level citability, Q&A formatting, entity clarity, structured data, and AI crawler accessibility. Generates citation capsules and a 0-100 AI Citation Readiness score. Use when user says "geo", "ai citation", "ai optimization", "citation audit", "aeo", "perplexity optimization", "chatgpt citation".

Analyzes and optimizes CLAUDE.md files following Anthropic's official best practices. Use when reviewing existing CLAUDE.md for improvements, or when user mentions CLAUDE.md is too long or ineffective.

Multi-agent consensus voting with domain-weighted expertise for critical decisions requiring structured validation

Token budgeting, placement effects, RAG patterns, prompt caching, compression, and multi-turn context strategies for LLM applications. Use when dealing with context windows, token budgets, retrieval-augmented generation, long context, context overflow, caching costs, observation masking, or compressing LLM inputs.

Generates production-grade Claude Code skills from unstructured brain dumps. Creates complete plugin packages with SKILL.md, agents, commands, and hooks. Use when the user asks to "create a skill", "build a skill", "add a skill", "make a skill that...", "write a skill for...", or "generate a skill". Do NOT use for general code generation, documentation writing, or modifying an existing skill — use improve-skill for that last case.

Creates Cursor-specific AI subagents with isolated context for complex multi-step workflows. Use when creating subagents for Cursor editor specifically, following Cursor's patterns and directories (.cursor/agents/). Triggers on "cursor subagent", "cursor agent". Do NOT use for generic subagent creation outside Cursor (use subagent-creator instead).

Manages toolsets for the Datadog MCP server `plugin:datadog:mcp`. Use when the user wants to view, enable, or disable toolsets that control which tools are available on the MCP server.

AI-native OCR platform that turns documents into high-accuracy data in minutes. Using multi-model consensus, DeepRead achieves 95%+ accuracy and flags only uncertain fields for review—reducing manual work from 100% to 5-10%. Zero prompt engineering required.

Interactive session to create Instructions field content for the Sanity Context MCP server. Use this skill whenever users mention tuning agent context, improving agent responses to Sanity data, configuring MCP instructions, setting up content filters, or when their agent gives wrong results from Sanity queries. Also trigger when users say their agent is confused about schema relationships, needs data-specific guidance, or wants to optimize which content the agent can access.

Access official collaboration platform documentation organized by platform components, protocols, and core competencies. Use when user asks about platform capabilities, how the framework works, or when Claude Code needs to reference architecture, implementation patterns, or best practices to answer questions accurately.

Vector embeddings configuration and semantic search

Explain a GitHub Pull Request (PR) or GitLab Merge Request (MR) to the user in plain, easy-to-understand language: WHAT was done, WHY/what for, and HOW — with the relevant code snippets embedded. Invoke this proactively and automatically right after creating or finishing a PR/MR (e.g. after running `gh pr create`, `glab mr create`, or pushing a branch and opening a PR/MR), even if the user did not explicitly ask for an explanation. Also use whenever the user asks to explain, summarize, walk through, recap, or 'tell me what you did' about a PR/MR or the changes in a branch. Works with any coding agent and relies on the local git diff, so it does NOT require gh/glab to function. Do NOT use for unrelated code reviews, bug hunting, or writing the PR/MR description itself — this skill only explains finished work back to the user.

On-device LLM integration using Apple's Foundation Models framework. Use when implementing AI text generation, structured output, or tool calling.

AI citability scoring and optimization. Analyzes web page content to determine how likely AI systems (ChatGPT, Claude, Perplexity, Gemini) are to cite or quote passages from the page. Provides a citability score (0-100) with specific rewrite suggestions.

Configure des hooks Claude Code pour bloquer les commandes git dangereuses (push, force-push, reset --hard, clean, branch -D, checkout/restore) avant leur exécution. Empêche les opérations git destructrices au niveau de l'agent.

Your intelligent GitHub command center -- start here. GitHub Hub discovers your repos and organizations, understands what you want to accomplish in plain English, and guides you to the right outcome by orchestrating every other agent. No commands to memorize. Just talk.

Home Assistant automations, scripts, and blueprints. Use when the user requests automation/script syntax, triggers/conditions/actions, blueprint usage, run modes, or troubleshooting, and when mapping those needs to ha-mcp automation/script tools.

Analyzes and optimizes existing Claude Code skills for better trigger precision, structural compliance, instruction quality, and agent efficiency. Produces measurable before/after scorecards. Use when the user asks to "improve a skill", "optimize a skill", "review a skill", "audit skill quality", "fix skill triggers", or references an existing SKILL.md that needs work. Do NOT use for creating new skills (use create-skill instead) or for reviewing non-skill code (use code-review instead).

Orchestrate the full Amazon KDP publishing workflow: audit, listing craft, manuscript prep, cover generation, dashboard submission, pricing, publish. Phases gate each other. Handles both eBook and paperback. Manages MCP-driven cover generation or falls back to manual prep.

Implements stateful agent graphs using LangGraph. Use when building graphs, adding nodes/edges, defining state schemas, implementing checkpointing, handling interrupts, or creating multi-agent systems with LangGraph.

Use this skill to research markets, understand trends, find opportunities, and validate ideas. Triggers: "market research", "market analysis", "market opportunity", "market size", "TAM SAM SOM", "industry analysis", "market trends", "validate idea", "market validation", "customer research", "market sizing", "opportunity analysis" Outputs: Market report with size, trends, opportunities, competitive landscape.

Your intelligent GitHub command center -- start here. Nexus discovers your repos and organizations, understands what you want to accomplish in plain English, and guides you to the right outcome by orchestrating every other agent. No commands to memorize. Just talk.

Set up OpenClaw locally and run it reliably on a Mac mini for private, always-on local agent workflows.

Multi-agent orchestration patterns for delegation, coordination, and conflict resolution

Divide-and-conquer implementation from specs/plans. Decomposes a reference document into independent tasks, assigns each to a builder agent, executes in parallel waves respecting dependencies, then integrates results. Use when you have a spec, PRD, plan, or large feature to implement quickly with parallel execution.

Defend AI systems against prompt injection and indirect prompt attacks using input controls, tool permissions, output validation, and isolation boundaries.

Pre-ship audit checklist for Ethereum dApps built with Scaffold-ETH 2. Give this to a separate reviewer agent (or fresh context) AFTER the build is complete. Use this skill whenever you are finalizing a dApp built with Scaffold-ETH 2.

Use when context window exhaustion is a concern or when addressing a large volume of PR comments — spawns a fresh agent session per comment via an external shell loop. For an in-session approach, use pr-review-loop or copilot-review-loop instead.

Use when experiments complete to judge what claims the results support, what they don't, and what evidence is still missing. Codex MCP evaluates results against intended claims and routes to next action (pivot, supplement, or confirm). Use after experiments finish — before writing the paper or running ablations.

Detects common LLM coding agent artifacts across four categories (tests, dead code, abstraction, style) over the project or changed files — using parallel subagents when the agent supports them, otherwise four sequential passes. Scans files changed since main by default; use --all for full-project scan. Triggers on LLM cruft cleanup, agent-generated code review, dead code sweeps, test-quality passes, or when the user asks to scan the whole repo.

Navigates role transitions for MCP work items using advance_item. Shows current role, gate status, required notes, and the correct trigger to use. Use when a user says: advance this item, move to work, start this task, complete this item, what's the next status, why can't I advance, unblock this, cancel this item, or check gate status.

Coordinate agent teams for adversarial review (paired skeptic/verifier per facet, challenge round with disposition vocabulary, consolidated findings with confidence) or parallel implementation (task sizing 5-6 per teammate, non-overlapping files). Enforces independent analysis before shared conclusions. Reference only (`disable-model-invocation: true`); loaded only when `agentTeams: true` in settings.

Load all project context for a GitLab issue — conventions, tech decisions, tech stack, defaults — confirm completeness with the developer, then enter Claude Code's native plan mode which plans and implements the solution. Detects existing work (branch, plan file) and offers to resume. Use when planning a ticket, saying "plan this ticket", "how should we build this", "implement this issue", or thinking through the approach. Run this after fetch-issue.

Load all project context for a Jira issue — conventions, tech decisions, tech stack, defaults — confirm completeness with the developer, then enter Claude Code's native plan mode which plans and implements the solution. Detects existing work (branch, plan file) and offers to resume. Use when planning a ticket, saying "plan this ticket", "how should we build this", "implement this issue", or thinking through the approach. Run this after fetch-issue.

Upload local files or folders into Tiangong KB through the Tiangong AI CLI. Use when a user asks an agent to ingest documents into a KB collection, list uploadable collections, or explain ingest duplicate/status results.

Analyze Trigger.dev tasks, schedules, and runs for cost optimization opportunities. Use when asked to reduce spend, optimize costs, audit usage, right-size machines, or review task efficiency. Requires Trigger.dev MCP tools for run analysis.

Searches for notes in the Carta CRM and helps the user update deal comments. Use this skill when the user says things like "update a note", "edit note", "update note content", "change a note", or "/update-note". Note: standalone note editing is not available via MCP — notes/comments are attached to deals and updated via the update-deal skill.

# System Architect Agent

Adds one or more company records to the Carta CRM via the Carta CRM MCP Server. Use this skill when the user says things like "add a company", "create company record", "add company to CRM", "add company to Carta CRM", or "/add-company". Collects company information conversationally, then creates it via the MCP server.

Stop waiting for prompts. Keep working.

Query the AIBTC agent network registry — look up agents by address or name, view network-wide stats, and rank agents by check-ins, achievements, or level.

Prepare and submit an MCP server to Anthropic's Connectors Directory. Use this skill whenever the user wants to list their MCP server on Claude's connector marketplace, submit to Anthropic's directory, get their MCP server featured in Claude Desktop or claude.ai, or prepare a connector submission. Also triggers on: 'submit to Anthropic', 'Claude connector directory', 'list on Claude marketplace', 'Anthropic MCP submission', 'get listed in Claude connectors', 'MCP directory submission form'. Even if the user just says 'how do I get my MCP server in the Claude directory' or 'I want Claude users to find my server', use this skill.

Onchain agent messaging on Base - post to feeds, send DMs, explore other agents

Bumps plugin versions across the Agent Alchemy ecosystem

Scan Claude Code changelog for new versions, features, and changes. Fetches official docs, parses release notes, and generates structured update report.

Scaffold cross-tool repo-local skills and agent instructions with canonical source in .agents/ and symlinks for Claude, Codex, Gemini, and Codex-home

Manage Coder workspaces and AI coding agent tasks via CLI. List, create, start, stop, and delete workspaces. SSH into workspaces to run commands. Create and monitor AI coding tasks with Claude Code, Aider, or other agents.

Threads the four-phase grain-to-coliseum transformation — given a dense prompt (the "grain"), autonomously surfaces parallel dimensions, composes self-contained 9-section handoff envelopes, dispatches in parallel, and reconciles returns into a composed whole under a mechanical compression gate. Use when a small-surface prompt plainly implies large scope; when the user invokes "grain → coliseum," "autonomous work assignment," "no ping-pong," or "chunk and hand off." Declines protocol when the grain is not fan-out shaped (≥3 independent dimensions).

Monitor community feedback on Claude Code via Exa search. Gathers sentiment, real-world usage patterns, bug reports, and feature requests from blogs, forums, and social media.

Create and configure configs in LaunchDarkly. Helps you choose between agent vs completion mode, create the config, add variations with models and prompts, and verify the setup.

Audit and rewrite AGENTS.md, CLAUDE.md, and other AI agent context files to be minimal and effective. This skill should be used when creating, reviewing, or rewriting context files (AGENTS.md, CLAUDE.md, .cursorrules, etc.) to follow research-backed guidelines that improve agent performance and reduce costs. Based on findings from arxiv.org/abs/2602.11988.

Configures or troubleshoots the Datadog MCP server `plugin:datadog:mcp`. Use when the user wants to change the Datadog domain, switch organizations, or when the server was previously configured but is not responding.

FastMCP TypeScript framework patterns for MCP servers. Auto-loads when building MCP servers, creating tools/resources/prompts, implementing authentication, configuring transports, or working with FastMCP in TypeScript.

This skill should be used when setting up fast file suggestions for Claude Code using ripgrep, jq, and fzf, or when improving file autocomplete performance or adding custom file suggestion behavior.

Open source project to make working with geospatial data in python easier. Extends the datatypes used by pandas to allow spatial operations on geometric types. Built on top of Shapely, Fiona, and Pyproj. Use for reading and writing spatial formats (Shapefile, GeoJSON, GeoPackage, KML), performing spatial joins, coordinate system transformations (reprojecting), geometric analysis (buffers, centroids, convex hulls), thematic mapping (Choropleth maps), calculating spatial relationships (contains, overlaps, touches, within), working with OpenStreetMap data or satellite-derived vector data.

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.

Session continuity for Claude Code. Gather context at start, archive state at end. Use when user mentions handoff, saving progress, or resuming work.

Persistent task ledger for agent coordination. Plan multi-step work, checkpoint progress across session boundaries, and coordinate across multiple agents with project pool routing.

Kai — Self-Improving Meta-Agent that detects recurring patterns in the file-based learnings store (.aidevteam/learnings/, written by /retro) and proposes permanent SKILL.md updates for human review. Clusters by target skill + theme; the Qdrant learnings/agent-knowledge collections are an optional overlay.

Behavioral rules and lifecycle event handlers for the core-tools package. Defines automated behaviors that trigger at specific points in the agent workflow. (converted from hooks)

Gateway to the Machina Sports premium platform — packaged agent workflows ("templates"), licensed real-time data, betting odds, and zero-latency live streams. This skill is prompt-only: it shells out to the separate `machina-cli` binary and routes the agent to a per-project Machina MCP server. Use when: the user asks for live odds, real-time telemetry, zero-latency match states, sub-second tick streams, packaged sports workflows (e.g., "Build a Bundesliga podcast bot", "Create a Polymarket arbitrage engine"), or when the open-source sports-skills are rate-limited or insufficient for the task. Don't use when: the user wants snapshot data from public APIs — use the sport-specific skill (nfl-data, polymarket, markets, …). Don't use to fetch data through raw HTTP — use the Machina MCP server, not a `requests` call.

Comprehensive guide for managing Claude Code snippets v2.0 - discovering locations, creating snippets from files, searching by name/pattern/description, and validating configurations. Use this skill when users want to create, search, or manage snippet configurations in their Claude Code environment. Updated for LLM-friendly interface with TTY auto-detection.

MCP architecture patterns, security, and memory management. Auto-loads when building MCP servers, implementing tools/resources, discussing MCP security, or working with FastMCP.

Use this skill when comparing AI or LLM models on benchmarks, capability, cost, latency, context window, or task-specific fit to help teams select the right model for their use case and budget. Not for training or fine-tuning models. Not for building eval frameworks from scratch.

Classify or explain Momentic test run results using Momentic MCP tools. Use when the user asks to categorize a failure, understand why a run failed, triage test results, or compare run results to past run results.

Guide users through configuring key MongoDB MCP server options. Use this skill when a user has the MongoDB MCP server installed but hasn't configured the required environment variables, or when they ask about connecting to MongoDB/Atlas and don't have the credentials set up.

Recursively initialize AGENTS.md in monorepo subdirectories with smart detection. Creates hierarchical agent context files with proper linking to root CLAUDE.md and parent AGENTS.md. Use for setting up multi-package projects, microservices, or any project with important subdirectories that need AI agent guidance.

Monitor LLMs and agentic apps: performance, token/cost, response quality, and workflow orchestration. Use when the user asks about LLM monitoring, GenAI observability, or AI cost/quality.

OKX OnChainOS: on-chain trading, analytics, security, DeFi, bridging across 20+ chains. Use when running OKX-routed on-chain ops (e.g. swap on Ethereum, scan token risk, track smart money, check wallet portfolio). Wallet: default to the user's Agent Wallet (via the `wallet` skill). Only use the OnchainOS TEE wallet (`onchainos wallet login <email>`) when the user explicitly asks for it.

Use the Paper MCP server to fetch paper metadata, sections, figures, citations, and full-text context, then translate that context into structured outputs such as summaries, extraction tables, and implementation-ready notes. Trigger when a task involves paper links/DOIs/arXiv IDs, literature review, citation extraction, or Paper MCP setup and troubleshooting.