Unified Minions skill for both deterministic shell jobs and LLM subagent orchestration. Replaces the older `gbrain-jobs` routing intent. Use when: submitting gbrain jobs, shell/background tasks, spawning subagents, checking progress, steering running work, pausing/resuming, parallel fan-out. One durable, observable, steerable queue interface.

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.

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Search personal knowledge bases, notes, docs, and meeting transcripts locally using qmd — a hybrid retrieval engine with BM25, vector search, and LLM reranking. Supports CLI and MCP integration.

Converts Product Requirements Documents into prd.json format for the Ralph autonomous agent system. Use when preparing PRDs for Ralph execution, breaking down features into atomic user stories, or when the user mentions Ralph, prd.json, or autonomous agent workflows.

Manage hierarchical task lists using the rune CLI tool. Create, update, and organize tasks with phases, subtasks, status tracking, task dependencies, and work streams for multi-agent parallel execution.

Use when delegating a task to an agent inside Sandvault macOS sandbox. Returns agent output (JSONL/JSON) for parsing.

Provides a clickable Airtable link whenever the agent has touched user-visible Airtable content. Use after every MCP call that creates, updates, lists, searches, or returns records, schema, or interface pages — bases, tables, fields, records, or pages. Hand off the most-specific URL the agent's tool calls have proven access to — prefer single-record URLs over table URLs, table URLs over base URLs, and interface page URLs when the user's access is restricted to pages. Format as a markdown link with a descriptive label. Construct URLs only from IDs the tools actually returned — never synthesize IDs to round out a URL. Compose this skill from any workflow skill that affects Airtable content.

Implement efficient similarity search with vector databases. Use when building semantic search, implementing nearest neighbor queries, or optimizing retrieval performance.

Implement efficient similarity search with vector databases. Use when building semantic search, implementing nearest neighbor queries, or optimizing retrieval performance.

The meta skill. Turn any raw feature into a properly-skilled, tested, resolvable unit of agent capability. Cross-modal eval is the recommended Phase 3 quality gate: 3 frontier models from different providers critique the output, you iterate to quality, THEN write tests that lock in the proven-good behavior.

Review Claude Code sub-agent implementations for best practices in configuration, tool access, hooks, and delegation patterns. Use when creating, auditing, or optimizing sub-agents.

Autonomous multi-agent task orchestration with dependency analysis, parallel tmux/Codex execution, and self-healing heartbeat monitoring. Use for large projects with multiple issues/tasks that need coordinated parallel execution.

Run a sprint retrospective for a completed Jira sprint. Gathers sprint data via the retro-analyzer agent, facilitates the retrospective conversation, captures team insights, and writes the retrospective report to Confluence. Use when running a retrospective, saying "retro", "what went well", "what didn't go well", reviewing how a sprint went, or reflecting on completed work. Also run when the sprint just ended and the team wants to capture lessons learned.

LaTeX TikZ/PGF package for programmatic vector graphics and diagrams. Use when helping users draw flowcharts, trees, graphs, automata, circuits, geometric figures, or any custom diagram in LaTeX.

Creates professional TikZ flowcharts with standardized themes, including Google Material-like and Anthropic-inspired options.

Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.

Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.

Triage issues through a state machine driven by triage roles. Use when user wants to create an issue, triage issues, review incoming bugs or feature requests, prepare issues for an AFK agent, or manage issue workflow.

Tool usage requirements, failure patterns, consistency checks, and validation methodologies for Claude Code operations

Use this agent for thinking through problems, exploring what your notes say, discussing where projects stand, and synthesizing across areas. A research partner that reads your vault and thinks with you. Writes synthesis to inbox.

Create or update multiple CMS items in a Webflow collection with validation and diff preview. Use when adding multiple blog posts, products, or updating fields across many items.

8-phase task execution lifecycle for teamwork workers using native Claude Code APIs. Covers task discovery, claiming, structured description parsing, TDD workflow, implementation, verification, selective commit, and completion reporting.

Use for multi-subagent parallel development. A main Claude instance manages multiple concurrent subagents, each working in isolated worktrees. Invoke with "/worktrees:orchestrator <feature>" or when user mentions "orchestrate agents", "parallel subagents", "coordinate development", "multi-agent workflow", "spawn subagents", or "divide feature into tasks".

Troubleshoot ZPA App Connector issues including enrollment failures, upgrade problems, Public Service Edge connectivity, and high CPU/memory/disk utilization. Uses MCP tools to inspect connector groups, provisioning keys, server groups, and application segments, then provides runbook-guided remediation steps. Use when an administrator reports 'connector is down', 'connector not enrolling', 'connector upgrade failed', or 'connector high CPU.'

INVOKE THIS SKILL when setting up a new project or when asked about package versions, installation, or dependency management for LangChain, LangGraph, LangSmith, or Deep Agents. Covers required packages, minimum versions, environment requirements, versioning best practices, and common community tool packages for both Python and TypeScript.

View recent Claude Code sessions sorted by last modified date. Shows a numbered list of sessions across all projects with summary, project, and message count.

Bootstrap agentic development environment from agent.toml manifest

Browser automation for web testing, form filling, screenshots, and data extraction. Ref-based workflow with best practices for reliable automation.

Build automated evaluation suites for AI agents using golden datasets, rubrics, and regression gates. Use when shipping agent features, validating prompt changes, or gating deployments on quality.

Re-analyze Claude Code session logs against existing agent guardrail rules to measure effectiveness, catch false positives, identify missed anti-patterns, and refine regex patterns. Use when user asks to "update guardrails", "refine guardrail rules", "check guardrail effectiveness", "agent-guardrails update", "tune guardrails", or wants to improve existing behavioral rules based on real usage data.

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

Manages and orchestrates prompts in Agent Platform. Use when you need to create, list, retrieve, version, or delete managed prompts in Agent Platform. Don't use for model training, model deployment to endpoints, or managing non-Agent Platform prompts.

Config-driven MCP (Model Context Protocol) server integration. Add external tool servers (GitHub, Brave Search, databases, etc.) by editing one config file — tools are auto-discovered, prefixed, and merged into the chat. Use this skill when the user says "add MCP", "connect MCP servers", "add external tools", or "setup ai-mcp".

AI/ML security playbook. Use when assessing model supply chain attacks (pickle RCE, poisoned weights), adversarial examples, model poisoning, model stealing, data privacy attacks (membership inference, model inversion), and autonomous agent security risks.

This skill should be used when the user asks to "start the angreal mcp server", "expose angreal tasks to Claude / Cursor / an AI agent", "configure .mcp.json for angreal", "set up MCP for my angreal project", "connect angreal to an AI assistant", "what does angreal mcp do", "AI agent integration with angreal", or needs guidance on the built-in Model Context Protocol server (`angreal mcp`), what it exposes, how MCP clients should be configured to connect to it, and how `ToolDescription` and `risk_level` flow into agent context.

Consult OpenAI Codex for investigation, debugging, or code review. Use when user explicitly asks to "ask codex", "check with codex", "codex review", or as a last resort when stuck after 4+ failed attempts at debugging, investigation, or bug fix and completely out of ideas. Codex is slow (2-5 min), so only escalate when truly stuck. Codex runs in read-only mode with full project access — it analyzes, we implement.

Use this agent for closed-loop test debugging - automatically analyzes test failures, suggests fixes, and re-runs tests until passing.

Automate legal client intake and onboarding with Box MCP — review uploaded intake documents for completeness against firm requirements, assess risk levels based on client profile and document content (PEP status, conflicts, sanctions, litigation history), route incomplete or high-risk submissions to appropriate attorneys with context and risk summaries, extract structured metadata (client name, matter type, jurisdiction, value), and generate engagement letters from Box DocGen templates for approved low-risk clients. Use this skill when the user mentions client intake, client onboarding, new client review, intake documents, engagement letters, or needs to process prospective client submissions stored in Box.

Guide for optimizing MSBuild build parallelism and multi-project scheduling. Only activate in MSBuild/.NET build context. USE FOR: builds not utilizing all CPU cores, speeding up multi-project solutions, evaluating graph build mode (/graph), build time not improving with -m flag, understanding project dependency topology. Note: /maxcpucount default is 1 (sequential) — always use -m for parallel builds. Covers /maxcpucount, graph build for better scheduling and isolation, BuildInParallel on MSBuild task, reducing unnecessary ProjectReferences, solution filters (.slnf) for building subsets. DO NOT USE FOR: single-project builds, incremental build issues (use incremental-build), compilation slowness within a project (use build-perf-diagnostics), non-MSBuild build systems. INVOKES: binlog MCP server tools (expensive_projects, expensive_targets, project_target_times); falls back to dotnet build -m, dotnet build /graph, binlog replay + grep.

Audit Claude Code setup for cache bugs (CC#40524): sentinel, --resume/--continue, attribution header + ArkNill B3/B4/B5

Guide for creating custom agents for Claude Code. Use when creating specialized agents, configuring agent tools.

Consult Claude specialist agents during implementation for codebase understanding, pattern checking, security review, debugging help, and more. Use this skill whenever you're unsure about conventions, stuck on a failure, or need expert input before writing code. Does not replace the formal review gates in agent-loops — this is for mid-implementation consultation.

Configure, manage, update and review Claude Code permissions, sandboxing, and tool access. Use when user wants to set up permissions, configure sandboxing, update allowed tools, manage settings.json permissions, or review permissions in skills or commands or agents or settings.json. When user writes a new skill, command, agent, or updates settings.json, they should use this skill to manage permissions.

ClawGang social skill — lets your agent socialize on clawgang.ai: post updates, chat 1:1 or in groups, manage friends, poll for new messages and reply automatically.

Maintains project context through progressive disclosure and enforced documentation. This skill should be used when starting work on a new project to initialize context documentation, or continuously during development to maintain up-to-date project structure and coding conventions. Triggers on project initialization, file modifications, or when the agent needs to understand project architecture.

This skill should be used when the user wants to create a PRD, write a product spec, plan a new product, define an MVP, scope out an idea, document product requirements, prepare specifications for AI coding agents, or says things like "I have an idea for an app," "help me plan my product," "write me a spec," or "turn my idea into a plan." It guides non-technical founders and technical builders through an adaptive brainstorming conversation that produces a comprehensive, AI-agent-optimized PRD in markdown compatible with any AI building tool.

Use when integrating detective skills across plugins. Maps agent roles to appropriate detective skills (developer → developer-detective, architect → architect-detective). Reference this to connect agents with claudemem investigation capabilities.

Review current diff and leave comments using diffity agent commands

Promote Doppel world builds across social platforms. Use when the agent wants to share builds on Twitter/X, Farcaster, Telegram, or Moltbook to drive observers, grow reputation, and recruit collaborators.

Generating or modifying .NET code. Common agent mistakes: async, NuGet, deprecated APIs, DI.

Two-party posture — user as director, agent as executor; every fork, tradeoff, or choice surfaced via batched AskUserQuestion with a recommended default. Use when the user invokes /duet, says "ask before" / "pair with me" / "human-in-the-loop", or for aesthetic/architectural/irreversible decisions.

Register AI agent identity on Ethereum via ERC-8004 Trustless Agents standard

Turn a refined research proposal or method idea into a detailed, claim-driven experiment roadmap. Use after `research-refine`, or when the user asks for a detailed experiment plan, ablation matrix, evaluation protocol, run order, compute budget, or paper-ready validation that supports the core problem, novelty, simplicity, and any LLM / VLM / Diffusion / RL-based contribution.

Understand how unfamiliar code works before planning or changing it — its structure, behavior, data flow, and the blast radius of a change. Use when the user says "explore", "investigate", "how does X work", "why does X happen", "where is X handled", "what calls X", "what would it take to change X", or whenever you need to understand code before acting on it. Drives exploration with the code_context MCP tool — symbol search, callgraph traversal, and blast-radius analysis — instead of reading files top to bottom.

Firecrawl CLI for web scraping, crawling, and search. Scrape single pages or entire websites, map site URLs, and search the web with full content extraction. Returns clean markdown optimized for LLM context. Use for research, documentation extraction, competitive intelligence, and content monitoring.

Applies fixes from a prior review-llm-artifacts run, with safe/risky classification. Respects verify-llm-artifacts output when present to skip false positives.

Query Google Gemini CLI configuration, features, and documentation. Use this skill proactively when the conversation involves: - Gemini CLI installation, setup, or authentication - Gemini CLI configuration (GEMINI.md, settings, themes) - Gemini CLI commands, flags, slash commands - Gemini CLI features (sandbox, plan mode, yolo, skills, hooks) - Gemini CLI model routing, model steering - Gemini CLI extensions, MCP, ACP mode - Gemini CLI headless/non-interactive mode - Gemini CLI policy engine, trusted folders - Gemini CLI session management, checkpointing, rewind

**The primary skill for building Glide apps.** Covers the full workflow, agent coordination, and browser automation. Use when building, creating, or modifying any Glide app. Read this first for any Glide project.

Use when the user wants to turn a rough idea, vague task, feature wish, bug-fix intent, or Codex /goal objective into a clear verifiable goal prompt. Use when the user mentions goal mode, /goal, set_goal, long-running autonomous Codex work, or asks to be interviewed/grilled before creating a goal.

Comprehensive interface for the Grandmaster AI chess platform. Play games, submit moves, and monitor matches.

Prompt engineering architect that transforms raw prompts into optimized, production-ready prompts. PROACTIVELY activate for: (1) Improve/refine prompts, (2) Optimize prompts for specific models, (3) Create prompt chains/sequences, (4) Analyze prompt quality, (5) Transform prompts for different contexts. Triggers: "improve this prompt", "refine prompt", "optimize prompt", "make this prompt better", "prompt engineering", "fix my prompt"

Manage Jira issues via Atlassian MCP — search, create, update, transition status, and handle sprint tasks. Auto-detects workspace configuration. Use when user says "create a Jira ticket", "update my sprint", "check Jira status", "transition this issue", "search Jira", or "move ticket to done". Do NOT use for Confluence pages (use confluence-assistant).

Pitfalls, examples, and edge cases for Hermes Kanban workers. The lifecycle itself is auto-injected into every worker's system prompt as KANBAN_GUIDANCE (from agent/prompt_builder.py); this skill is what you load when you want deeper detail on specific scenarios.

INVOKE THIS SKILL when setting up a new project or when asked about package versions, installation, or dependency management for LangChain, LangGraph, LangSmith, or Deep Agents. Covers required packages, minimum versions, environment requirements, versioning best practices, and common community tool packages for both Python and TypeScript.

Launch an intelligent sub-agent with automatic model selection based on task complexity, specialized agent matching, Zero-shot CoT reasoning, and mandatory self-critique verification

Use this agent when you need to search institutional learnings in docs/solutions/ for relevant past solutions before implementing a new feature or fixing a problem. This agent efficiently filters documented solutions by frontmatter metadata (tags, category, module, symptoms) to find applicable patterns, gotchas, and lessons learned. The agent excels at preventing repeated mistakes by surfacing relevant institutional knowledge before work begins.\n\n<example>Context: User is about to implement a feature involving email processing.\nuser: "I need to add email threading to the brief system"\nassistant: "I'll use the learnings-researcher agent to check docs/solutions/ for any relevant learnings about email processing or brief system implementations."\n<commentary>Since the user is implementing a feature in a documented domain, use the learnings-researcher agent to surface relevant past solutions before starting work.</commentary></example>\n\n<example>Context: User is debugging a performance issue.\nuser: "Bri...

Compose Mapbox MCP tools to produce grounded, cited location-aware responses from live data instead of training data

Internal helper agent. Invoked by orchestrator agents via Task tool. Internal helper for scanning a single markdown file for accessibility issues across all 9 domains. Returns structured findings with severity, line numbers, suggested fixes, and auto-fix classification. Invoked by markdown-a11y-assistant via the Task tool - not user-invokable directly.

Model Context Protocol server management and tool integration

Production-ready patterns and best practices for MCP servers - architecture, security, performance, and maintenance

Parks the current Claude Code session before going to sleep. Analyzes the conversation, writes a handoff document to docs/handoffs/ in the current project, mirrors the same summary to engram tagged with the session ID, and prints the exact `claude --resume <id>` command for tomorrow. Zero interaction — runs end-to-end on a single invocation. Trigger: /night-park, night park, park session, me voy a dormir, guardar sesion.

Reference for nimble agent commands. Load for Step 0 agent lookup. Contains: full agent table (50+ sites across e-commerce, food, real estate, jobs, social, travel), discover/list/schema/run commands, response shapes (PDP=dict, SERP=list, google=entities), agent memory.

Use at the start of any OpenCLI session — this is the top-level map of what `opencli` can do, how to discover adapters, what flags and output formats are universal, and which specialized skill to load next. Point here when an agent asks "what can opencli do?" or "how do I find the right command?".

Scaffold the .orchestra/ agent knowledge base in a project — creates folder structure, templates, README, and ADR-000.

Techniques for getting varied, non-repetitive outputs from repeated LLM calls. Use when building systems that call LLMs in loops, generating personality lines, commentary, names, or any repeated creative text where outputs feel samey or cliche.

Gemini vision-powered PDF to markdown converter. Handles scanned docs, multi-column layouts, tables, footnotes, flowcharts, and degraded documents that text-based extraction destroys. Uses two-tier model routing (cheap model for clean digital pages, capable model for everything else) with per-chunk confidence scoring, anti-hallucination detection, and continual learning from user corrections. Use when a user needs to convert, extract, analyze, or process any PDF document — especially scanned documents, government forms, legal contracts, academic papers, or anything where pypdf/pdfplumber returns garbage or nothing.

Create complete Claude Code plugins with proper structure including skills, commands, agents, and hooks. Generates plugin manifests and directory structures for distribution. Use when: building plugins, creating plugin packages, distributing skills to marketplace. Activates for: "create plugin", "generate plugin", "new plugin", "plugin package", "make plugin"

Use this skill when the user asks to "find duplicate issues", "clean up duplicates", "what dupes are in the backlog", "/pm-dedup", or after a /pm-status briefing flags likely-dup clusters. Runs the `duplicates` lens against the team cache, has the agent judge each cluster, then offers merge / relate / close per cluster. Powered by the shared dedup engine in `hooks/lib/dedup.js`.

Complete AI coding workflow system. Orchestration patterns, 18 hook events, 5 agents, cross-agent support, reference guides, and searchable learnings. Works with Claude Code, Cursor, and 32+ agents.

Use this skill when you writing commands, hooks, skills for Agent, or prompts for sub agents or any other LLM interaction, including optimizing prompts, improving LLM outputs, or designing production prompt templates.

Read an existing research outline and fan out independent background agents to deeply research each item, producing one structured JSON per item against the shared field schema. Resumable, batched, with output disabled per agent (each agent has its explicit output file). Use when `outline.yaml` + `fields.yaml` from `/research-outline` are in place and you want to parallelise the per-item investigation — academic paper deep-dives, product comparisons, benchmark fills, multi-target competitive analysis. Trigger this after `/research-outline`, before `/research-report`.

Multi-agent adversarial verification with convergence loop. Two independent review agents must both pass before output ships.

Guides an MCP work item through its schema-defined lifecycle — filling required notes using guidancePointer and advancing through gate-enforced phases. Internal skill triggered by hooks and output styles during orchestration workflows. Use when an item has schema tags and needs to progress through queue, work, review, or terminal phases with note gates.

Search Claude Code sessions by keyword across all projects. Searches summaries and first prompts by default, or full transcripts with --deep.

Synchronize project CLAUDE.md with recent codebase changes by analyzing git history, reviewing against official Anthropic best practices using parallel agents, and proposing comprehensive updates. Use when CLAUDE.md is outdated or doesn't exist.

Run a sprint retrospective for a completed GitLab milestone. Gathers sprint data via the retro-analyzer agent, facilitates the retrospective conversation, captures team insights, and writes the retrospective report to Google Drive. Use when running a retrospective, saying "retro", "what went well", "what didn't go well", reviewing how a sprint went, or reflecting on completed work. Also run when the sprint just ended and the team wants to capture lessons learned.

Show enriched daily activity from Claude Code sessions — branches, PRs, Linear issues, and what you worked on. Use when the user wants to see what they did today, review their daily work, or get a summary of their sessions.

Adversarial test verification. Proves AL test suites are meaningful by mutation sweeps, assertion auditing, and coverage gap analysis. Always runs as a sub-agent with clean context — never inline.

Canonical front door for agent-first workload planning, runtime mode selection, workflow-rig dogfood, and runtime dispatch in joelclaw. Use when the user says 'run this through the workflow rig', 'kick this off', 'dogfood this with the workflow rig', 'start a canary', 'run this as a workload', or any request to turn coding/runtime intent into a real joelclaw workload plan or run.

Troubleshoot Zscaler Microsegmentation (ZMS) agent deployment and connectivity issues. Investigates agent fleet health, connection status, version compliance, agent group configuration, provisioning keys, and TOTP secrets. Use when an administrator reports: 'Agents are disconnected', 'Agent enrollment failing', 'How do I provision new agents?', 'Check agent versions', or 'Agent not connecting.'

# DevOps & Deployment Engineer Agent

Automate ActiveCampaign tasks via Rube MCP (Composio): manage contacts, tags, list subscriptions, automation enrollment, and tasks. Always search tools first for current schemas.

Background knowledge for droid-control workflows -- not invoked directly. Agent-browser driver mechanics for web page and Electron desktop app automation.

Build, validate, and troubleshoot deep links for Codex, Cursor, VS Code, Visual Studio, and similar tools. Use when users ask for clickable links (especially in Slack) that open threads, files, folders, or app settings.

Audit and reduce AI agent token and inference spend through context discipline, prompt caching, model routing, batching, and workflow capture. Use when discussing AI coding bills, token waste, model selection, prompt caching, or agent cost optimization.