LLM Mart Basic

Use when the user asks to "test the marketplace plugins", "run marketplace tests", "check every installed plugin", or "/ar:marketplace-test". Walks the installed marketplace plugins, finds each one's test suite, runs them from the plugin's own directory, and reports per-plugin re

Render Mermaid diagrams to SVG using beautiful-mermaid and bun. Use when the user asks to "render mermaid", "generate diagram", "create flowchart", "update diagrams", "render SVG from mermaid", "beautiful-mermaid", "regenerate diagrams", or needs to convert Mermaid diagram syntax

Eliminate a stubborn configuration bug by testing 8-10 fundamentally different fixes in parallel instead of one at a time. Use when the root cause is ambiguous and asked to "fix this stubborn bug", "we've tried everything", "run approaches in parallel", "use parallel subagents",

Use when the user asks for the "system design philosophy", "design principles", "/ar:philosophy", or "/ar:ph", and before major feature work, an architecture decision, or a pull-request review. Lists the 17 universal system design principles, ordered from most fundamental to most

Use when the user asks to "make a plan", "plan this out", "/ar:plannew", or "/ar:pn", or starts multi-step work that needs tracked steps. Creates a structured plan with checkbox steps, the evidence each step must produce, and the wait process between them.

Use when the user asks to "execute the plan", "run the plan", "/ar:planprocess", or "/ar:pp". Executes an approved plan step by step with the wait process, task tracking, and verification before each completion claim.

Use when the user asks to "refine the plan", "critique the plan", "/ar:planrefine", or "/ar:pr". Critiques an existing plan against the actual code, records a change ledger, and iterates until a full pass finds no new material issue.

Use when the user asks to "update the plan", "sync the plan with the code", "/ar:planupdate", or "/ar:pu". Reconciles an existing plan with the current codebase and marks what already shipped.

Edit prose for brevity and scanability without losing facts. Use for requests to "revise", "rewrite", "edit", "polish", "clean up", "clarify", "simplify", "condense", "tighten", "shorten", "streamline", "de-fluff", remove bloat/repetition or unnecessary newlines, fix spelling/typ

Use when the user asks to "list my claude sessions", "which windows are waiting", "/ar:tabs". Discovers Claude sessions across tmux windows and reports which need attention.

Use when the user asks to "continue every session", "send this to all windows", "/ar:tabw". Acts on Claude sessions across tmux windows; destructive to in-flight work, so it states what it will do before doing it.

Drive CLI tests inside isolated tmux/byobu sessions with ai-monitor integration. Use when asked to "test this CLI", "run it in tmux", "automate a terminal session", "capture the output of an interactive command", "send keystrokes to a session", or to exercise a plugin or terminal

autorun control commands, stated as prose after $ar: status; allow/justify/find file-creation policy; ok/no/blocks/clear command guards; global variants; stop/estop; task tracking and pause; cache-miss gate; planexport settings; reload; help lists everything.

This skill should be used when the user asks to "extract text from PDF", "convert PDF to text", "parse PDF", "read PDF contents", "extract data from documents", "batch PDF extraction", "PDF to markdown", "OCR PDF", "get text from PDF files", "I have a PDF", "can you read this PDF

Expertise in maintaining, debugging, and deploying the autorun hook system across Claude Code, Codex CLI, Gemini-family CLIs, Google Antigravity, Qwen Code, ForgeCode, custom harnesses, and desktop app integrations. Use when the user asks to "fix hooks", "deploy autorun", "debug

Hugging Face Inference SDK patterns for TypeScript/Node.js — InferenceClient setup, chat completion, text generation, streaming, embeddings, image generation, audio transcription, translation, summarization, and Inference Endpoints

Replicate SDK patterns for TypeScript/Node.js -- client setup, predictions, streaming, webhooks, file handling, model versioning, deployments, and training

Together AI SDK patterns for TypeScript — client setup, chat completions, streaming, structured output, function calling, embeddings, image generation, fine-tuning, and OpenAI-compatible endpoints

LLM observability with Langfuse — OpenTelemetry-based tracing, evaluations, prompt management, datasets, and production best practices

Testing and evaluation framework for LLM prompts and applications -- promptfooconfig.yaml, assertions, model-graded evals, red teaming, CI/CD integration, custom providers, and comparative evaluation

Fourteen posts of being wrong in production, compressed to checkboxes

Healthy nodes, a quiet network, 300 restarts in three days, and a latency budget measured in milliseconds

Discovery worked. Ping worked. Every TCP connection timed out, and later the tunnel only worked when someone had a terminal open.

Every VM came back. The cluster did not. Declarative systems converge on config, and the datapath isn't config.

A surprising share of AI-in-the-terminal failures aren't the AI. They're zsh, and a version of bash from 2006.

A Claude Code plugin turns standalone project configuration into a namespaced, installable extension that teams and communities can update as one unit.

None of the safety came from the model. It came from six boring habits.

Skills package instructions and references. Subagents run work in a separate context and return results. They solve different problems and can be composed deliberately.

Six hours in, one step left, everything green, and the incident that didn't happen

CLAUDE.md carries persistent project context. Skills load reusable procedures when relevant. Separating stable facts from task-specific workflows keeps both easier to maintain.

Twenty minutes recovering secrets that never existed, and the one sentence from a human that ended it

An API request routing a model's tool call through an approval gate to a remote MCP server

31 config keys, two audits, and why the first one was wrong in both directions

The official MCP Registry stores standardized server metadata rather than package code. Publishers verify a namespace, describe installation or remote access, and submit immutable versions.

Everyone looks at the Dockerfile. The file that actually leaked the key was the project file.

Remote MCP authorization uses established OAuth standards, but secure integration still requires issuer validation, least-privilege scopes, protected token handling, and server-side enforcement.

"Copy it over and switch the reference" is two steps, and the outage lives in the one nobody checks

stdio fits local processes and prototypes. Streamable HTTP fits hosted services and shared integrations. The right choice follows where the capability runs and who must reach it.

The most important rule wasn't about what I could change. It was about what I was allowed to display.

Tools perform operations, resources expose readable context, and prompts provide reusable templates. Choosing the correct primitive makes an MCP server easier to understand and govern.

Turn a GAIA workflow misfire into a redacted, classified, filing-ready bug report in one read-only pass. Self-diagnoses config issues inline; files probable bugs upstream on confirmation.

Judge-the-form, human-gated hardening. Reviews recurring code-audit-frontend findings and, with approval, drafts the lowest-context-weight form (deterministic check / skill / path-scoped prose rule) into the working tree. Pass `list` to see live candidates or `why <finding_class>` to explain one.

Initialize a new project from the GAIA React template, renames, strips GAIA branding, configures i18n, installs Claude skills/plugins.

Plan a complex feature using GAIA's task-orchestration pattern, structures the work into fresh-context subagent phases for your approval. Does not implement.

Cut a new GAIA release, bump version, graduate CHANGELOG, regenerate manifest, open release PR, then tag on merge. Maintainer-only.

Detect and, on explicit consent, additively append the languages Serena is not indexing to the `languages:` list in `.serena/project.yml`, then prompt a Serena restart. Never mutates without a yes; inert without Serena.

Author an immutable SPEC artifact through Socratic discovery (spec-kit wrapper), then STOP. Terminal, never runs /gaia-plan; it prints a /gaia-plan prompt the human pastes into a fresh session. Pass `auto <description>` for non-interactive mode that answers its own questions.

Maintainer-only autonomous health audit + auto-heal loop. Runs N=3 fresh-team audit-fix-audit cycles with circuit breakers, reports an F-to-A+ verdict (folding in the shared Claude-integration fitness grade) or escalates.

Single post-init onboarding command; detects situation, runs only owed phases; safe to re-run. --reconfigure rotates token and re-selects tools.

GAIA before_specify hook: constitution placeholder check + spec-kit version-pin drift detection.

GAIA after_specify hook: immutability lint over the just-written SPEC artifact.

Close a plan after implementation+merge. Offers wiki-promote for the plan's consolidated SUMMARY.md, cold-consolidates an out-of-band merge, then early-reaps the local plan folder once cost is represented in cost.jsonl.

GAIA self-review: pre-gate-2 review pass on the in-progress SPEC draft.

Close a SPEC after implementation+merge. Optional drain of deferred wiki-promote, cold-consolidates an out-of-band merge into SUMMARY.md, then early-reaps the local SPEC folder once cost is represented in cost.jsonl.

GAIA Socratic discovery wrapper: /speckit-specify for the initial draft, then GAIA's own Socratic clarify loop.

GAIA before_implement hook: render PO-authored UATs into Playwright e2e specs at .playwright/e2e/spec-NNN/.

Promote merged SPEC or plan content into the GAIA wiki.

This project uses the GAIA preset. Bare `/speckit-clarify` is not the clarify path here: core clarify writes an off-shape artifact (a `## Clarifications` / `### Session` block with five-word answers) and carries a question cap GAIA does not use. Run `/gaia-spec` instead — it driv

GAIA-wrapped /speckit-specify: writes through core, then relocates the artifact to .gaia/local/specs/SPEC-NNN/SPEC.md and stamps GAIA frontmatter.

Show or configure the compact Fallow Impact statusline in Claude Code

Claude Code skills for Chinese-narration recaps from supported video files, with optional editable JianYing/CapCut draft export. | 用 Claude Code skills 为支持的视频文件…

Drive the Unity Editor from an AI agent or the terminal. The Editor serves MCP itself over HTTP, so there is no second process to run, and the isuzu-unity-cli c…

面向企业的数字员工构建与运行平台：把专业员工的经验、流程与判断标准，固化为可随时上岗、可配置、可审批、可观测的 AI 数字员工。

A local-first, Pi-powered AI agent workspace for Desktop, WebUI, and CLI

660+ muapi-hosted generative-media models plus community-submitted third-party API tools (SEO, enrichment, social, scraping) — one YAML file per entry, browsabl…

A compounding agent OS for recursive agents. Also an open source alternative to Grok Bot and Meta's Muse.

Independent desktop client for OpenCode 2. Manage projects, sessions, parallel agents, requests, and changes on Linux and macOS.

Production agent skills for Claude Code, Cursor, and any SKILL.md harness — Codex fleets, video pipeline, monorepo review bundles, multi-chain explorer.

See and manage what your coding assistants load — skills, commands, subagents, plugins and MCP servers, with real usage.

An Enterprise-Grade Full-Stack RBAC Permission Management System Built with Go + React

Create AI Agents in a No-Code Visual Builder or TypeScript SDK with full 2-way sync. For shipping AI assistants and multi-agent AI workflows.

基于 DeepSeek Harness（DSH）的稳定桌面端，集成Git、内置浏览器与记忆功能 | DeepSeek Harness desktop GUI with local workspaces, Git, browser and memory.

⚡ Control Apache Airflow with natural language via MCP. Chat with your workflows using Claude, GPT, or any LLM — no REST API calls needed. Supports Airflow 2.x…

Open-source CLI, schemas, resolver, and DSH agent tools for DSH Plugin Hub

Pi Coding Agent 中文学习蓝皮书：从安装与第一个可验收任务开始，逐步掌握 Session、Context、Skill、Extension、Subagent 与长期 Agent 工作流。

Self-hosted framework for orchestrating fleets of specialist AI agents — ensemble reasoning and a full agentic coding pipeline, model-agnostic and local-friendl…

Self-hosted, vendor-neutral control plane for your local coding agents (Claude Code & Codex). Run your agents, on any plane.

A practical AI agents handbook covering agent systems, agentic workflows, LangGraph, MCP/A2A, context engineering, agent memory, evaluation, observability, and…

Vibe Coding 从入门到精通教程｜AI 结对编程工作流｜Prompt、Skill、Workflow、上下文管理、codex实战指南

Static security scanner for LLM agents — prompt injection, MCP config auditing, taint analysis. 51 rules mapped to OWASP Agentic Top 10 (2026). Works with LangC…