All news with #cloudflare tag

🔎 Cloudflare launches isitagentready.com and a companion Cloudflare Radar dataset to measure and accelerate adoption of emerging AI agent standards across the web. The tool scores sites on Discoverability, Content, Bot Access Control, and Capabilities, and returns actionable prompts for each failing check. The site publishes machine-readable endpoints (MCP server, agent-skills index) so compatible agents can scan and remediate programmatically. Cloudflare also refactored its developer docs to serve Markdown and curated LLM resources, producing measurable reductions in token usage and latency.

📦 Cloudflare is introducing support for shared compression dictionaries to reduce redundant transfers and speed page loads for sites that deploy frequently or are heavily crawled by agents. In Phase 1 the edge will passthrough Use-As-Dictionary and Available-Dictionary headers and respect dcb/dcz encodings; an open beta begins April 30, 2026. Later phases move delta compression and automatic dictionary generation into Cloudflare’s edge, simplifying origin logic and maximizing bandwidth and latency savings for versioned assets and returning visitors.

🚀Using Real User Measurements that capture browser‑side timing via a small background speed test, Cloudflare reports it became the fastest provider in 60% of the top 1,000 networks by December 2025, up from 40% in September. Rankings rely on the trimean of TCP connection time to smooth outliers and reflect real user experience. Improvements came from new points of presence (Wroclaw, Malang, Constantine) and software optimizations such as HTTP/3 support and tighter congestion handling, producing an average 6ms lead over the next provider in December.

🔁 Cloudflare introduces Redirects for AI Training, a toggle that turns existing rel="canonical" tags into HTTP 301 redirects for verified AI training crawlers. On paid Cloudflare plans this enforcement redirects AI crawler traffic (examples include GPTBot, ClaudeBot, Bytespider) to canonical URLs, preventing ingestion of deprecated content. Human visitors and other automated classes are unaffected.

💾 Cloudflare's Unweight is a lossless compression system for LLM weights that reduces model size by roughly 15–22% while preserving bit-exact outputs and requiring no special hardware. It compresses only the exponent byte of BF16 tensors—using Huffman coding, palette/transcoding and row-level fallbacks—while leaving sign and mantissa untouched. Decompression happens into GPU shared memory to feed tensor cores directly, and Cloudflare has published a technical paper and open-sourced GPU kernels.

🧠 Cloudflare announces Agent Memory, a private beta managed service that extracts information from agent conversations and makes it available without filling model context windows. The service offers persistent profiles with operations to ingest conversations, explicitly remember or forget items, and recall synthesized answers, integrating with Cloudflare Workers and a REST API. Agent Memory uses a retrieval-based architecture with deterministic ingestion, multi-stage verification, vector and full-text retrieval channels, and Reciprocal Rank Fusion to synthesize concise, contextual responses. Memories are classified, versioned or superseded as appropriate, and fully exportable so organizations retain ownership.

🧭 Cloudflare introduces Flagship, a native feature-flag service built on the CNCF standard OpenFeature that evaluates flags at the edge using Workers, Durable Objects, and KV. The Worker binding performs in-isolate evaluations with typed accessors and full evaluation details, avoiding external HTTP calls and reducing latency. Flagship centralizes flag storage, change auditing, percentage rollouts, and nested targeting rules, and is now available in private beta to help teams safely ship autonomous or AI-assisted code.

🚀 Cloudflare details optimizations to run extra-large open-source LLMs on Workers AI, notably making Kimi K2.5 three times faster and adding more models. The post explains hardware tuning, prefill–decode disaggregation, token-aware load balancing, and prompt-caching via an x-session-affinity header to improve throughput and tail latency. It also covers KV-cache sharing with Mooncake, speculative decoding with NVIDIA EAGLE-3, and Cloudflare’s Rust-based inference engine Infire for multi-GPU, low-memory, fast cold-start inference.

🚀 Cloudflare announced a unified AI inference platform that lets developers call models from multiple providers through one API and a single billing plane. Using the same AI.run() binding for Workers, teams can switch providers with a one-line change; REST support is coming for other environments. The update adds default gateways, automatic retries, finer logging, and centralized cost monitoring to reduce latency, improve reliability, and control costs for agentic workflows.

🔗 Cloudflare now enables developers to create and manage PlanetScale Postgres and MySQL databases directly from the Cloudflare dashboard and API, with billing to your Cloudflare account launching next month. The integration uses Hyperdrive to provide managed connection pools and query caching so Workers can run fast, reliable SQL queries with standard Postgres clients. Configuration is simple: add a Hyperdrive binding to your Worker and connect using the connection string provided, allowing low-latency global access and familiar PlanetScale tooling.

🗂 Artifacts is a Git-compatible, versioned filesystem built for agent-first workflows. It enables programmatic repo creation, credential issuance, and commit operations via a REST API or a native Workers API while remaining accessible to any standard Git client. Cloudflare implements Artifacts on Durable Objects with a Zig-to-WASM Git engine and supports import, forking, git-notes, and session-scoped repositories. The feature is in private beta for paid Workers plans, with a public beta expected in early May.

🔍 Cloudflare introduced AI Search, a plug-and-play search primitive that provides a unified retrieval layer for agents, support bots, and coding assistants. It pairs hybrid semantic (vector) and BM25 keyword matching, managed storage, and built-in vector indexes so developers can create instances at runtime without provisioning separate infrastructure. The service integrates with Workers, the Agents SDK, and Wrangler, supports metadata boosts and cross-instance queries, and can optionally rerank results with a cross-encoder.

✉️ Cloudflare has opened Email Sending to public beta, enabling applications and agents to send transactional email directly from Workers or via REST and SDKs. The service pairs with Email Routing to provide full bidirectional workflows—receive, process, persist, and reply—without external infrastructure. Built-in deliverability features (SPF, DKIM, DMARC) and global delivery on Cloudflare’s network simplify operations. Tooling additions include a Workers binding, MCP server, Wrangler CLI commands, a Cloudflare Email skill, and an open-source Agentic Inbox reference app.

🤖 Agent Lee is an in-dashboard AI assistant that understands your Cloudflare account and can troubleshoot, visualize, and modify resources on demand. Built on Cloudflare primitives like Workers, Durable Objects, and the Agents SDK, it uses a TypeScript Codemode to generate server-side calls that run through a sandboxed MCP proxy. Write operations require explicit user approval via an elicitation gate, and the beta — serving ~18,000 daily users and ~250k tool calls per day — is available now.

🔎 Cloudflare has renamed Browser Rendering to Browser Run and expanded it into a remote browser platform tailored for AI agents. It runs full Chrome sessions across Cloudflare’s global network and can be driven by code or AI using Puppeteer, Playwright, Quick Actions, direct CDP access, or MCP/WebMCP. New features include Live View for real-time inspection, Human-in-the-Loop handoffs for operator intervention, a CDP endpoint for low-level automation, and Session Recordings for replay and debugging. The service raises concurrency and Quick Actions limits and is available on Workers Free and Paid plans.

🧠 Project Think is Cloudflare’s next-generation Agents SDK, introducing primitives and an opinionated base class to build long-running, serverless AI agents. It combines durable execution with fibers, isolated sub-agents (Facets) backed by per-agent SQLite, persistent tree-structured sessions, and sandboxed code execution via Dynamic Workers and @cloudflare/codemode. The Think base class wires workspace, toolchains, resumable streams, and extensions together so developers can deploy resumable, auditable agents that cost nothing when idle.

🚀 Cloudflare today launched the Registrar API in beta, enabling programmatic domain search, availability checks, and direct registration. Designed for editors, CI pipelines, and agent-driven workflows, the API exposes a simple Search → Check → Register flow that agents can use to suggest names, confirm pricing, and complete purchases using account defaults. Registrations are offered at-cost, with WHOIS privacy enabled by default and explicit fee acknowledgement required for premium domains.

🔊 Cloudflare released an experimental voice pipeline for the Agents SDK via the @cloudflare/voice package, enabling real-time voice input and output on the same agent architecture developers already use. The extension preserves the same Durable Object instance, Agent class, tooling, and SQLite-backed persistence while adding withVoice, withVoiceInput, React hooks, a framework-agnostic VoiceClient, and built-in Workers AI STT/TTS providers to get started without external API keys. Audio streams over the existing WebSocket, uses continuous STT for turn detection, and streams sentence-chunked TTS to improve time-to-first-audio, while offering hooks, provider adapters, telephony options, and support for multimodal text/voice interactions.

🚀 Cloudflare upgraded Workflows with V2, a rearchitected control plane to meet machine-speed, agent-driven workloads. The update raises defaults to 50,000 concurrent instances, 300 instances/sec per account, and 2,000,000 queued instances per workflow, and introduces horizontally scalable components SousChef and Gatekeeper to distribute metadata and concurrency slots. The redesign preserves reliability via Engine-driven instance state, Durable Object alarms, and a staged, zero‑downtime migration that converted legacy Account Durable Objects into SousChefs.

🔐 Cloudflare outlines how it centralized and secured company-wide use of the Model Context Protocol (MCP) by combining controls from Cloudflare One and its developer platform. The post explains why locally hosted MCP servers posed supply‑chain and administration risks and how Cloudflare moved to governed, remote MCP servers with Access-based authentication, audit logging, and CI/CD templates. It highlights MCP server portals, Code Mode to reduce token costs, and Gateway detection for shadow MCP.