LLM & Agents

Run LLMs and AI models on Cloudflare's global GPU network with Workers AI. Includes Llama 4, Gemma 3, Mistral 3.1, Flux image generation, BGE embeddings (2x faster, 2025), streaming support, and AI Gateway for cost tracking. Use when: implementing LLM inference, generating images, building RAG with embeddings, streaming AI responses, using AI Gateway, troubleshooting max_tokens defaults (breaking change 2025), BGE pooling parameter (not backwards compatible), or handling AI_ERROR, rate limits, model deprecations, token limits. Keywords: workers ai, cloudflare ai, ai bindings, llm workers, @cf/meta/llama-4-scout, @cf/google/gemma-3-12b-it, @cf/mistralai/mistral-small-3.1-24b-instruct, @cf/openai/gpt-oss-120b, workers ai models, ai inference, cloudflare llm, ai streaming, text generation ai, ai embeddings, bge pooling cls mean, image generation ai, workers ai rag, ai gateway, llama workers, flux image generation, deepgram aura, leonardo image generation, vision models ai, ai chat completion, AI_ERROR, rate limit ai, model not found, max_tokens breaking change, bge pooling backwards compatibility, model deprecations october 2025, token limit exceeded, neurons exceeded, workers ai hono, ai gateway workers, vercel ai sdk workers, openai compatible workers, workers ai vectorize, workers-ai-provider v2, ai sdk v5, lora adapters rank 32

Build AI applications with OpenAI Agents SDK - text agents, voice agents (realtime), multi-agent workflows with handoffs, tools with Zod schemas, input/output guardrails, structured outputs, and streaming. Deploy to Cloudflare Workers, Next.js, or React with human-in-the-loop patterns. Use when: building text-based agents with tools and Zod schemas, creating realtime voice agents with WebRTC/WebSocket, implementing multi-agent workflows with handoffs between specialists, setting up input/output guardrails for safety, requiring human approval for critical actions, streaming agent responses, deploying agents to Cloudflare Workers or Next.js, or troubleshooting Zod schema type errors, MCP tracing failures, infinite loops (MaxTurnsExceededError), tool call failures, schema mismatches, or voice agent handoff constraints.

Build with OpenAI's stateless APIs - Chat Completions (GPT-5.2, GPT-5.1, o3, o4-mini), Realtime API (voice), Batch API (50% savings), Embeddings, Images (DALL-E 3), Audio (Whisper + TTS), and Moderation. Node.js SDK and fetch for Cloudflare Workers. Use when: implementing chat with GPT-5.2/5.1/o3/o4-mini, Realtime voice API (WebSocket), Batch API for cost savings, xhigh reasoning (GPT-5.2), streaming responses, function calling/tools, structured outputs, embeddings for RAG, DALL-E 3 images, Whisper transcription, TTS (13 voices), or troubleshooting rate limits (429), API keys (401), streaming errors.

Add OAuth authentication to MCP servers on Cloudflare Workers. Uses @cloudflare/workers-oauth-provider with Google OAuth for Claude.ai-compatible authentication. Use when building MCP servers that need user authentication, implementing Dynamic Client Registration (DCR) for Claude.ai, or replacing static auth tokens with OAuth flows. Prevents CSRF vulnerabilities, state validation errors, and OAuth misconfiguration.

Generate hierarchical AGENTS.md structures for codebases. Use when user asks to create AGENTS.md files, analyze codebase for AI agent documentation, set up AI-friendly project documentation, or generate context files for AI coding assistants. Triggers on "create AGENTS.md", "generate agents", "analyze codebase for AI", "AI documentation setup", "hierarchical agents".

Build autonomous AI agents with Claude Agent SDK. Structured outputs (v0.1.45, Nov 2025) guarantee JSON schema validation, plugins system, hooks for event-driven workflows. Use when: building coding agents with validated JSON responses, SRE systems, security auditors, or troubleshooting CLI not found, structured output validation, session forking errors.

Build with Claude Messages API using structured outputs (v0.69.0+, Nov 2025) for guaranteed JSON schema validation. Covers prompt caching (90% savings), streaming SSE, tool use, model deprecations (3.5/3.7 retired Oct 2025). Use when: building chatbots/agents with validated JSON responses, or troubleshooting rate_limit_error, structured output validation, prompt caching not activating, streaming SSE parsing.

Build AI agents with Cloudflare Agents SDK on Workers + Durable Objects. Includes critical guidance on choosing between Agents SDK (infrastructure/state) vs AI SDK (simpler flows). Use when: deciding SDK choice, building WebSocket agents with state, RAG with Vectorize, MCP servers, multi-agent orchestration, or troubleshooting "Agent class must extend", "new_sqlite_classes", binding errors.

Build MCP servers in Python with FastMCP framework to expose tools, resources, and prompts to LLMs. Supports storage backends (memory/disk/Redis), middleware, OAuth Proxy, OpenAPI integration, and FastMCP Cloud deployment. Use when: creating MCP servers, defining tools or resources, implementing OAuth authentication, configuring storage backends for tokens/cache, adding middleware for logging/rate limiting, deploying to FastMCP Cloud, or troubleshooting module-level server, storage, lifespan, middleware order, circular imports, or OAuth errors.

Proactively analyze user requests at the start of conversations to determine task type, assess prompt quality, and intelligently recommend which skills to activate. Should activate for ALL user requests to ensure optimal workflow. Evaluates clarity, specificity, and completeness to suggest prompt-optimizer when needed. Identifies UI design tasks for ui-analyzer and component requests for react-component-generator. Acts as intelligent skill coordinator.

Comprehensive CrewAI framework guide for building collaborative AI agent teams and structured workflows. Use when developing multi-agent systems with CrewAI, creating autonomous AI crews, orchestrating flows, implementing agents with roles and tools, or building production-ready AI automation. Essential for developers building intelligent agent systems, task automation, and complex AI workflows.

Multi-session Claude workflow orchestration. Spawn workers via TabzChrome, coordinate parallel tasks, use subagents for monitoring/exploration, manage beads issues. Use this skill when coordinating multiple Claude sessions or managing complex multi-step workflows.

Quantum-resistant, self-learning version control for AI agents with ReasoningBank intelligence and multi-agent coordination

Focus testing effort on highest-risk areas using risk assessment and prioritization. Use when planning test strategy, allocating testing resources, or making coverage decisions.

Apply the Holistic Testing Model evolved with PACT (Proactive, Autonomous, Collaborative, Targeted) principles. Use when designing comprehensive test strategies for Classical, AI-assisted, Agent based, or Agentic Systems building quality into the team, or implementing whole-team quality practices.

AI agents as force multipliers for quality work. Core skill for all 19 QE agents using PACT principles.

API contract testing, authentication flows, rate limit handling, and error scenario coverage for n8n integrations with external services. Use when testing n8n node integrations.

Comprehensive mobile testing for iOS and Android platforms including gestures, sensors, permissions, device fragmentation, and performance. Use when testing native apps, hybrid apps, or mobile web, ensuring quality across 1000+ device variants.

Testing methodologies, test-driven development (TDD), unit and integration testing, and testing best practices across multiple frameworks. Use when the user needs to write tests, implement TDD, or improve test coverage and quality.

Searching internet for technical documentation using llms.txt standard, GitHub repositories via Repomix, and parallel exploration. Use when user needs: (1) Latest documentation for libraries/frameworks, (2) Documentation in llms.txt format, (3) GitHub repository analysis, (4) Documentation without direct llms.txt support, (5) Multiple documentation sources in parallel