AI Jury
Convene a panel of native coding-agent CLIs from **different vendors** to review the same change, debate each other's findings, and produce one consolidated verdict. This is multi-model review where each agent runs in its own native CLI (with its own tooling), not API-level prompting.
aws-cdk
Provides AWS CDK TypeScript patterns for defining, validating, and deploying AWS infrastructure as code. Use when creating CDK apps, stacks, and reusable constructs, modeling serverless or VPC-based architectures, applying IAM and encryption defaults, or testing and reviewing `cdk synth`, `cdk diff`, and `cdk deploy` changes. Triggers include "aws cdk typescript", "create cdk app", "cdk stack", "cdk construct", "cdk deploy", and "cdk test".
pydantic-evals
Test and evaluate AI agents and LLM outputs using code-first evaluation framework with strong typing. Use when the user wants to: (1) Create evaluation datasets with test cases for AI agents, (2) Define evaluators (deterministic, LLM-as-Judge, custom, or span-based), (3) Run evaluations and generate reports, (4) Compare model performance across experiments, (5) Integrate evaluations with Pydantic AI agents, (6) Set up observability with Logfire, (7) Generate test datasets using LLMs, (8) Implement regression testing for AI systems.
rails-audit-thoughtbot
Perform comprehensive code audits of Ruby on Rails applications based on thoughtbot best practices. Use this skill when the user requests a code audit, code review, quality assessment, or analysis of a Rails application. The skill analyzes the entire codebase focusing on testing practices (RSpec), security vulnerabilities, code design (skinny controllers, domain models, PORO with ActiveModel), Rails conventions, database optimization, and Ruby best practices. Outputs a detailed markdown audit report grouped by category (Testing, Security, Models, Controllers, Code Design, Views) with severity levels (Critical, High, Medium, Low) within each category.
product-architect
Complete product development system with 31 specialized agents and 23 frameworks. Use when user asks to build a product, write a PRD, create a roadmap, plan an MVP, design an app, do a security audit, create a financial model, plan hiring, launch a product, set up operations, prepare for IPO, or write a compliance policy. Also triggers on help me plan, product strategy, go-to-market, fundraising, pitch deck, unit economics, competitive analysis, user personas, sprint planning, SOP, checklist for, or how do I start a company. Do NOT use for general knowledge questions, coding tutorials, or creative writing unrelated to product development.
claw-orchestrator
Manage persistent coding sessions across Claude Code, Codex, Gemini, and Cursor engines. Use when orchestrating multi-engine coding agents, starting/sending/stopping sessions, running multi-agent council collaborations, cross-session messaging, ultraplan deep planning, ultrareview parallel code review, or switching models/tools at runtime. Triggers on "start a session", "send to session", "run council", "ultraplan", "ultrareview", "switch model", "multi-agent", "coding session", "session inbox", "cursor agent".
Inconvo CLI â Semantic Model Skill
Use this skill when you need to update an agent semantic model or user-context from code context. This repository uses a CLI-only mutation workflow.
lazyllm-skill
LazyLLM framework for building multi-agent AI applications. Use when task mentioned LazyLLM or AI program for: (1) Flow orchestration - linear, branching, parallel, loop workflows for complex data pipelines, (2) Model fine-tuning and acceleration - finetuning LLMs with LLaMA-Factory/Alpaca-LoRA/Collie and acceleration with vLLM/LMDeploy/LightLLM. Includes comprehensive code examples for all components, (3) RAG systems - knowledge-based QA with document retrieval, vectorization, and generation, (4) Agent development - single/multi-agent systems with tools, memory, planning, and web interfaces.
agents-consilium
Query external AI agents (Codex, Gemini, OpenCode, Claude Code headless) in parallel for independent second opinions, code review, bug investigation, and consensus on high-stakes decisions. Agents and models are configurable in config.json. Use for architecture choices, security review, or ambiguous problems where independent perspectives matter. Not for simple questions answerable from docs or the codebase â use web search or repo exploration instead.
Io.Github.Kevint Cerebras/Cerebras Code Mcp
Model Context Protocol (MCP) server for Cerebras to make coding faster in AI-first IDEs
Antigravity CLI (`agy`)
Google's agentic coding CLI. Runs Gemini 3 (Pro/Flash) by default, with internal access to Claude Sonnet/Opus 4.5/4.6 and GPT-OSS models. Pairs a chat REPL with planning, scheduling, and autonomous "goal" workflows that write `implementation_plan.md` + `task.md` to a per-conversation brain directory
Brownfield Discovery
Reverse-engineer an existing project to produce a Design Doc, API contracts, and a Threat Model anchored in the real code, IaC, and observability. Run once per project. Output anchors all subsequent /spec and /build invocations.
Agent Dev â Vault-First Internal Development
Develop the agent's own internals with the vault as the primary source of truth. The vault knows more about the agent than any code scan or model training data. Always search the vault first, extract maximum context, and only then touch code.
midnight-compact
Write, test, and deploy Compact smart contracts for Midnight. Use when writing privacy-preserving contracts, ZK circuits, shielded tokens, or any on-chain Midnight code. Triggers on: Midnight, Compact, smart contract, zero-knowledge, ZK, shielded, circuit, witness, ledger, proof server, DUST, NIGHT, disclose, Zswap. Covers Compact language syntax, privacy model, circuit patterns, testing, security best practices, SDK integration, and wallet connectivity.
fast-mlx
Optimize MLX code for performance and memory. Use when asked to implement or speed up MLX models or algorithms, reduce latency/throughput bottlenecks, tune lazy evaluation, type promotion, fast ops, compilation, memory use, or profiling.
aura
All-in-one fullstack dev engine for Claude Code. Build, fix, clean, deploy, review in one /aura command. Discovery-First · Tiered Model · 6-Layer Security · 33 Modes · 8 Languages · 23 Hooks · 75% Token Savings. Use when user asks to build, fix, clean, deploy, review, debug, qa, brainstorm, orchestrate, or create content.
adding-models
Guide for adding new LLM models to Letta Code. Use when the user wants to add support for a new model, needs to know valid model handles, or wants to update the model configuration. Covers models.json configuration, CI test matrix, and handle validation.
Unity Agent Harness for Claude Code — auto-compile, test pipelines, cross-model review, out of the box.
Compare the current branch against develop and generate two review documents: architecture change diagram + PR review checklist.
ai-pair
AI Pair Collaboration Skill. Coordinate multiple AI models to work together: one creates (Author/Developer), two others review (Codex + Gemini). Works for code, articles, video scripts, and any creative task. Trigger: /ai-pair, ai pair, dev-team, content-team, team-stop
Threat Modeling Skill v3.1.0 (20260313a)
AI-native automated software risk analysis skill. LLM-driven, Code-First approach for comprehensive security risk assessment, threat modeling, security testing, penetration testing, and compliance checking.
codex-delegate
Delegate a heavy or mechanical coding task to the configured Codex CLI model via a low-effort Sonnet wrapper, so Codex's large transcript stays OUT of the main orchestrator context. Trigger when the user says "delegate to codex", "hand this to codex", "let codex do X", or when a well-specified implementation/refactor/migration/bulk task is large enough that doing it inline would flood context with tool round-trips. Do NOT use for small tasks you can one-shot.
claude-plan â independent plan from Claude's plan mode
The host model planning its own work shares its own blind spots. A different-vendor model (Anthropic Claude, pinned to Fable) has a different training distribution, so where its plan diverges from yours is exactly where to dig. This skill runs the Claude Code CLI (`claude -p`) in Claude's **native p
laravel-best-practices
Apply this skill whenever writing, reviewing, or refactoring Laravel PHP code. This includes creating or modifying controllers, models, migrations, form requests, policies, jobs, scheduled commands, service classes, and Eloquent queries. Triggers for N+1 and query performance issues, caching strategies, authorization and security patterns, validation, error handling, queue and job configuration, route definitions, and architectural decisions. Also use for Laravel code reviews and refactoring existing Laravel code to follow best practices. Covers any task involving Laravel backend PHP code patterns.
ast-index
This skill should be used when the user asks to "find a class", "search for symbol", "find usages", "find implementations", "search codebase", "find file", "class hierarchy", "find callers", "module dependencies", "unused dependencies", "project map", "project conventions", "project structure", "what frameworks", "what architecture", "find Perl subs", "Perl exports", "find Python class", "Go struct", "Go interface", "find React component", "find TypeScript interface", "find Rust struct", "find Ruby class", "find C# controller", "find Dart class", "find Flutter widget", "find mixin", "find Scala trait", "find case class", "find object", "find PHP class", "find Laravel model", "find PHP trait", or needs fast code search in Android/Kotlin/Java, iOS/Swift/ObjC, Dart/Flutter, TypeScript/JavaScript, Rust, Ruby, C#, Scala, PHP, Perl, Python, Go, C++, or Protocol Buffers projects. Also triggered by mentions of "ast-index" CLI tool.
cobbler
Invoke Elves Cobbler, the default orchestration model for fitted answers and agent routing. Use when the user types /cobbler or asks for Cobbler coordination in Claude Code.
/interview-cheatsheet â long-form Chinese ML/LLM interview prep
Generate one comprehensive Chinese cheat sheet per invocation: formulas + derivations + from-scratch code + 25 é«é¢é¢. Output passes cross-model math/code review before rendering. **Detect-only by default: never auto-commits.**
Io.Github.Index9 Org/Mcp
Up-to-date model data for AI coding assistants.
Rails Backend Skill
Opinionated conventions for Rails backend code. Architecture is built on a single principle: **place the domain model at the center.** Controllers, jobs, and the console are all boundaries that orchestrate domain logic â they contain no business logic themselves.
Model Radar
Ping 134+ free coding LLMs across 17 providers, rank by latency, pick the fastest.
302ai-api-integration
ALWAYS use this skill when user needs ANY API functionality (AI models, image generation, video, audio, text processing, etc.). Automatically search 302.AI's 1400+ APIs and generate integration code. Use proactively whenever APIs or AI capabilities are mentioned.
cross-model-review
Use this skill for cross-model code reviews where a different AI model reviews code written by the current model. Activates on mentions of cross-model review, peer review, second opinion, review my code, review this PR, review changes, independent review, unbiased review, different model review, cross-check code, or code review.
direnv â Per-Project Claude Code Environments
Set up `.envrc` files so Claude Code automatically loads the right API keys, provider, model, and feature flags when you `cd` into a project directory.
modlens
Plug-in vision for text-only models. Hard rule: when a file path or URL with an image extension (.png, .jpg, .jpeg, .webp, .gif, .heic, .heif) appears anywhere in the conversation (typed by the user, injected as a `[Image: source: <path>]` line, or inside a tag) and you cannot see that image's content, run this skill on it before any other approach: no self-built OCR, no PIL, no tesseract. Also triggers on pasted-image placeholders such as `[Image #1]` and `[Unsupported Image]`. If you can actually see the image, do not use this skill. When unsure, run `modlens guard` before the first read of a session: a deny verdict means the active model has native vision and must read the image itself. Runs the modlens CLI to convert the image into structured JSON evidence: every word transcribed, layout regions, semantics, visual clues. Also use when the user asks how to install, configure, or switch modlens providers (Gemini API key, OpenAI-compatible endpoints, Claude API or Claude Code CLI).
Io.Github.Thezenmonster/Agentscore
KYA — verify any AI agent. Six checks: deployer, model, code, abuse, permissions, deployment.
gemini-interactions-api
Use this skill when writing code that calls the Gemini API for text generation, multi-turn chat, multimodal understanding, image generation, streaming responses, background research tasks, function calling, structured output, or migrating from the old generateContent API. This skill covers the Interactions API, the recommended way to use Gemini models and agents in Python and TypeScript.
Multi-MCP
Multi-model AI orchestration MCP server with code review, compare, and debate tools.
spatie-laravel-php-standards
Apply Spatie's Laravel and PHP coding standards for any task that creates, edits, reviews, refactors, or formats Laravel/PHP code or Blade templates; use for controllers, Eloquent models, routes, config, validation, migrations, tests, and related files to align with Laravel conventions and PSR-12.
Io.Github.Mikhae1/Kubeview
Read-only Model Context Protocol MCP server enabling code-driven AI analysis of Kubernetes clusters.
cost-mode
Cost-conscious Claude Code mode. Reduces token usage 40-70% by enforcing concise responses, smart model routing, and efficient workflow patterns. Keeps full technical accuracy. Activate with /cost-mode or "enable cost mode". Auto-triggers on mentions of budget, cost, tokens, or spending.
Vision for a text-only agent
Your model cannot receive image content directly. When the user pastes an image, Claude Code shows it as an `[Unsupported Image]` placeholder â a text marker with **no usable pixel data**. You cannot read it. Instead, obtain the image through a route you can access, then send it through the `analy
mechanical-engineering-research
Research, write, code, analyze, present, and develop proposals for thermal-fluid mechanical engineering work with source-aware rigor. Use for heat transfer, fluid mechanics, thermodynamics, HVAC, energy systems, turbomachinery, pumps, piping, CFD, experiments, correlations, standards, datasheets, papers, patents, AI/ML tools, computer vision, sequence regression, surrogate modeling, research coding, Overleaf, VS Code, GitHub, git, federal grant proposals, DOE/NSF/NASA-style narratives, invention disclosure, provisional patent support, commercialization, and trade studies. Produces technical briefs, critical literature reviews, proposal narratives, review-criteria responses, manuscript sections, methods, results discussions, data-analysis plans, plots, presentations, design comparisons, calculation plans, reproducible code, repository workflows, disclosure drafts, patent-support packets, and research roadmaps.
pinchbench
Run PinchBench benchmarks to evaluate OpenClaw agent performance across real-world tasks. Use when testing model capabilities, comparing models, submitting benchmark results to the leaderboard, or checking how well your OpenClaw setup handles calendar, email, research, coding, and multi-step workflows.
pupu-test-api
Use when running QA / regression tests against PuPu, when verifying a code change actually works in the running app, or when reading PuPu UI/state without screenshotting manually. Triggers on tasks like "test that PuPu still creates chats correctly", "verify the new model selector works end-to-end", "send a message and check the response", "what's PuPu's current state?". Phase 1 covers chat lifecycle, message send (blocking), model/toolkit/character switching, logs, state snapshot, screenshot, eval.
ai-cost-manager-reducer
Audit AI-powered features, API calls, agent workflows, prompt chains, tool calls, embeddings, vector search, retries, streaming flows, background jobs, cron tasks, caching layers, logs, and model selection logic to detect token waste, repeated AI calls, and inefficient prompt chains. Produces severity-ranked findings with safe optimization strategies that preserve reasoning quality and chain reliability. Use before approving or modifying any AI-powered feature, agent workflow, or model-calling code.
alignment-review
Review code changes for bugs and alignment with OpenEnv principles and RFCs. Use when reviewing PRs, checking code before commit, or when asked to review changes. Implements two-tier review model.
synapptic
synapptic â The missing feedback loop for agentic development. Builds a living user model from AI coding session transcripts. Extracts user preferences, AI failure patterns, and behavioral guards, then writes them to the memory system so every new session starts informed. Use this skill whenever the user mentions: synapptic, user profile, user archetype, session analysis, extract observations, update profile, profiling sessions, user model, guards, known weaknesses, AI failures, behavioral rules, or wants to analyze how they work with their AI coding assistant. Also trigger when the user asks to 'process sessions', 'update the archetype', 'show my profile', 'what do you know about me', 'run synapptic', or wants to improve the AI's understanding of their preferences.
mythic-engineering
Activate Mythic Engineering mode â architecture-first, document-guided, AI-orchestrated development with six specialist roles. Invoke at the start of any session or task to orient the session, load the correct role, and enforce the full ME protocol including the five-layer model, coding laws, daily routine, and required checklists.
api-docs-generator
Audits and enhances FastAPI and REST API documentation: missing descriptions, response codes, examples, docstrings, Pydantic models, OpenAPI spec. Triggers on: "generate API docs", "document this API", "OpenAPI for", "FastAPI docs", "document endpoints", "swagger docs".