cloud▌

Run LLMs, embeddings, and image generation on Cloudflare's GPU network with 14 new 2025 models, streaming support, and 7 documented error preventions. \n \n Supports 40+ models across text generation (Llama 4, Gemma 3, Mistral 3.1, GPT-OSS), embeddings (BGE 2x faster, EmbeddingGemma), image generation (Flux, Leonardo), vision, and audio (Deepgram, Whisper v3) \n Handles critical 2025 breaking changes: context window validation switched from characters to tokens, BGE pooling parameter no longer b

OAuth authentication for MCP servers on Cloudflare Workers with Google Sign-In and Dynamic Client Registration. \n \n Implements dual OAuth role pattern: MCP server acts as both OAuth client (to Google) and OAuth server (to MCP clients like Claude.ai), issuing its own tokens after upstream authentication \n Includes production-ready security: CSRF protection via HttpOnly cookies, one-time-use state tokens with 10-minute TTL, session binding via SHA-256 hashing, and HMAC-signed approval cookies t

Durable, long-running workflows with automatic retries, state persistence, and event handling on Cloudflare Workers. \n \n Supports three core step types: step.do() for work execution with configurable retries, step.sleep() for delays, and step.waitForEvent() for external event handling with timeouts \n Automatically persists serializable state across steps and hibernation; prevents 12 documented errors including waitForEvent timeout bugs, I/O context violations, and non-idempotent operation pit

Category: test

Build Python serverless APIs on Cloudflare Workers with async-only execution, external package support, and multi-step workflow automation. \n \n Deploy Python APIs using the WorkerEntrypoint class pattern with pywrangler CLI; supports all Cloudflare bindings (D1, KV, R2, Workers AI, Durable Objects, Queues) \n Requires async-only code: use httpx or aiohttp for HTTP calls, avoid sync libraries like requests and native C extensions \n Python Workflows enable durable multi-step DAG automation with

Stateful Durable Objects for real-time apps, coordination, and persistent state management. \n \n SQLite-backed storage with 10GB capacity, SQL queries, and atomic transactions; KV backend available for simpler key-value use cases \n WebSocket hibernation API supports thousands of concurrent connections with automatic wake-up; message size limit increased to 32 MiB (Oct 2025) \n Alarms API for scheduling future tasks, batching, and cleanup without relying on setTimeout / setInterval \n RPC metho

Backend service development with long connections, multi-language support, and AI agent capabilities. \n \n Supports two modes: Function mode (Node.js, built-in HTTP/WebSocket/SSE, fixed port 3000, local running) and Container mode (any language/runtime via Docker, no local tool support) \n Requires stateless services that listen on the PORT environment variable and write data externally; resource constraints enforce Mem = 2 × CPU \n Includes read tools ( queryCloudRun ) for listing services and

Manage Twill Cloud Coding Agent workflows through the public v1 API. \n \n Create, list, retrieve, and manage tasks with support for custom branches, agent selection, and file attachments \n Stream job logs in real time via Server-Sent Events and cancel running jobs at any point \n Full lifecycle control: send follow-up messages, approve plans, archive tasks, and manage task state \n Schedule recurring coding tasks with cron expressions, timezone support, and pause/resume capabilities \n List re

Four MCP tools for safe SQL execution and security rule management on CloudBase Relational Database. \n \n Four tools cover all agent operations: executeReadOnlySQL for SELECT queries, executeWriteSQL for INSERT/UPDATE/DELETE/DDL, readSecurityRule to inspect permissions, and writeSecurityRule to update access controls \n Use MCP tools exclusively in agent context; do not initialize CloudBase SDKs when these tools are available \n Always run read-only SELECTs first to verify assumptions before de

Architects design workloads by integrating platform services, functionality, and code to meet both functional and nonfunctional requirements. To design effective workloads, you must understand these requirements and select topologies and methodologies that address the challenges of your workload's constraints. Cloud design patterns provide solutions to many common challenges.