[
{"id":"cf3fa7dee91fd91d3274109d6efb48737fc3f63d09729a5d41449be338637b15","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"1c7a6e8038b8bde722496eabd41a55b1f8a4086109b26c540012937c00ddc138","change_id":"1c7a6e8038b8bde722496eabd41a55b1f8a4086109b26c540012937c00ddc138","provider":"cloudflare","verdict":"minor","rationale":"The entry adds third-party models to the existing Workers AI catalog, not a new product or capability.","facts":"{\"title\":\"Workers AI - DeepSeek V4 Flash and Pro now available on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-08-14-deepseek-v4-workers-ai/\",\"summary\":\"@cf/deepseek-ai/deepseek-v4-pro-0813 and @cf/deepseek-ai/deepseek-v4-flash-0731 are now available on Workers AI. DeepSeek V4 Flash and DeepSeek V4 Pro are the first Workers AI models with a full one million (1,048,576) token context window . Use them for long-horizon agentic workflows, large codebases, and multi-step reasoning that exceed the context limits of every other model hosted on the platf\",\"date\":\"2026-08-14\"}","detected_at":"2026-08-14","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"2d7055fa832fa1d432d66533301f28bc11a74d68b11a9932a11a61324a5731eb","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"6c5499fc53a7f93a769026e532facec4895695df91c8684120d9f024dbc26a87","change_id":"6c5499fc53a7f93a769026e532facec4895695df91c8684120d9f024dbc26a87","provider":"cloudflare","verdict":"minor","rationale":"The entry adds a single third-party model to the existing Workers AI catalog.","facts":"{\"title\":\"Workers AI - Qwen 3.8 27B now available on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-08-17-qwen-3.8-27b-workers-ai/\",\"summary\":\"@cf/qwen/qwen3.8-27b is now available on Workers AI. Qwen 3.8 27B is a 27-billion-parameter instruction-tuned vision language model from Alibaba's Qwen family. It processes images and text together, with reasoning and function calling for agentic workflows. Key capabilities: Vision : Accept image and text inputs and generate text responses. Reasoning : Support thinking mode for complex, step-by-st\",\"date\":\"2026-08-17\"}","detected_at":"2026-08-17","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"8c3eeb955f3f3fd529cf78b73666092b354b7d0a1e3997e0b92da955803ec538","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"fa0075bc797bc39da224ea3fc93bc0b31e635926bbf7dc49245980e06da5d067","change_id":"fa0075bc797bc39da224ea3fc93bc0b31e635926bbf7dc49245980e06da5d067","provider":"cloudflare","verdict":"minor","rationale":"The entry makes OAuth scopes optional during authorization, a small option.","facts":"{\"title\":\"Agents, Workers - Choose OAuth scopes for Wrangler and the Cloudflare API MCP server\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-08-22-wrangler-mcp-optional-oauth-scopes/\",\"summary\":\"Wrangler and the Cloudflare API MCP server now use optional OAuth scopes. During authorization, you can choose which optional scopes to grant instead of approving every scope requested by each client. The consent dialog now includes the option to edit the permissions you grant to Wrangler or the Cloudflare API MCP server: You can then choose which specific permissions to grant: Required scopes rem\",\"date\":\"2026-08-22\"}","detected_at":"2026-08-22","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"9dbe2675349350ddd37a8b4d697fade1435c1e971891fa7a2b6f3e5fea7d26a5","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"d3ae541bde77ee7b2f5380987f5e3996f2d6df49ea3d8423a7b7b8567d2cbc50","change_id":"d3ae541bde77ee7b2f5380987f5e3996f2d6df49ea3d8423a7b7b8567d2cbc50","provider":"cloudflare","verdict":"minor","rationale":"The entry adds a single third-party model to the existing Workers AI catalog.","facts":"{\"title\":\"Workers AI - Z.ai GLM-5.3 Flash now available on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-08-26-glm-5.3-flash-workers-ai/\",\"summary\":\"@cf/zai-org/glm-5.3-flash is now available on Workers AI. It is the first natively multimodal model in the GLM-5 series, built on a Mixture-of-Experts architecture with 320B total parameters and 18B active per token. GLM-5.3 Flash is the first GLM-family model on Workers AI to support multimodal inputs. It outperforms GLM-5.2 across benchmarks and real-world workloads at a lower price, while appro\",\"date\":\"2026-08-26\"}","detected_at":"2026-08-26","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"de27893847850839cea6849245671ea6a2db113876f62150b9ecafa4f1f4a7f4","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"82365e000c8b66f53b1d8e327ccc48d4453ca161836deb9d2a1f9df9315f777d","change_id":"82365e000c8b66f53b1d8e327ccc48d4453ca161836deb9d2a1f9df9315f777d","provider":"cloudflare","verdict":"minor","rationale":"The entry adds a single third-party model to the existing Workers AI catalog.","facts":"{\"title\":\"Workers AI - Z.ai GLM-5.3 now available on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-08-28-glm-5.3-workers-ai/\",\"summary\":\"@cf/zai-org/glm-5.3 is now available on Workers AI. It is Z.ai's flagship agentic coding model, built for long-running, tool-driven development workflows rather than single-turn chat. GLM-5.3 uses the same base model as GLM-5.2, with every gain coming from post-training. The results are substantial on coding and agentic benchmarks: Z.ai reports ↗︎ a 50% improvement over GLM-5.2 on its in-house Z.a\",\"date\":\"2026-08-28\"}","detected_at":"2026-08-28","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"add475ec90def40656e34cf9c5195b686e6831d94f3957bbb08595eb22b7aee5","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"afbadb5d19629fad3b977df5c6b5bbe0ac3431f5dca3b574ea3208e479915871","change_id":"afbadb5d19629fad3b977df5c6b5bbe0ac3431f5dca3b574ea3208e479915871","provider":"cloudflare","verdict":"minor","rationale":"The entry adds turn metrics to the existing voice package, a small option rather than a new product.","facts":"{\"title\":\"Agents - Inspect Voice Agent turn latency and outcomes\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-09-11-voice-diagnostics-turn-metrics/\",\"summary\":\"@cloudflare/voice v0.4.0 now lets you inspect where each Voice Agent turn spends time and how it ends. client. addEventListener ( \\\"turnmetrics\\\" , ( turn ) => { console. log (turn.outcome, turn.turnTotalMs); }); About the Voice package The @cloudflare/voice package lets you build real-time voice agents with Cloudflare Agents. It streams microphone audio to an Agent over WebSocket, transcribes speec\",\"date\":\"2026-09-11\"}","detected_at":"2026-09-11","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"0186b1654564f97d0722d07f85e7684b360ab95b3e30fe25afb0589791a31bc2","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"b6e9c517096d9a77d2bcc37b961df6798e978b55eaddadf632a638b795c8c7d6","change_id":"b6e9c517096d9a77d2bcc37b961df6798e978b55eaddadf632a638b795c8c7d6","provider":"cloudflare","verdict":"minor","rationale":"The entry adds the rejectIfBusy option to inference requests, a small option.","facts":"{\"title\":\"Workers AI - Reject busy synchronous inference requests\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-09-17-reject-if-busy/\",\"summary\":\"The rejectIfBusy option lets synchronous Workers AI inference requests fail when capacity is unavailable. Use it when your application should not wait in a capacity queue. Pass the option as the third argument to the Workers AI binding: const response = await env. AI . run ( \\\"@cf/google/gemma-4-26b-a4b-it\\\" , { messages: [{ role: \\\"user\\\" , content: \\\"Explain capacity queues.\\\" }], }, { rejectIfBusy: t\",\"date\":\"2026-09-17\"}","detected_at":"2026-09-17","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"1cc5b57218fd2be02b9b43a52bfb3850501d9e735589b2610a162a6e4169b70f","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"bb6d22ddfce0013057f74404f5822c3fff9a3820fee39dfcf04bbf56e6a4d319","change_id":"bb6d22ddfce0013057f74404f5822c3fff9a3820fee39dfcf04bbf56e6a4d319","provider":"cloudflare","verdict":"major","rationale":"The entry says the Workers OAuth Provider is now v1 with a new split API, a new major version of an SDK.","facts":"{\"title\":\"Agents, Workers - The best way to do MCP auth just got better: Workers OAuth Provider goes v1, with a new split API and full support for MCP 2026-07-28\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-01-workers-oauth-provider-1x/\",\"summary\":\"@cloudflare/workers-oauth-provider ↗︎ is now v1, with a new split API. One Worker acts as the authorization server: it signs users in and issues tokens. Your MCP server acts as the resource server, and can run in another Worker. It validates each token with the authorization server over a Service Binding , without crossing the public Internet. It supports the MCP 2026-07-28 authorization specifica\",\"date\":\"2026-10-01\"}","detected_at":"2026-10-01","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"5ca2813f25acc3f8ef2e88060fbd7499d22d3879ca5c82044c9f6ad1f2612c9c","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"68491f535d68320a856ca54b3724379117aaa6bbe1b29da271966a2a28f935bf","change_id":"68491f535d68320a856ca54b3724379117aaa6bbe1b29da271966a2a28f935bf","provider":"cloudflare","verdict":"major","rationale":"The entry introduces Clef, the first models trained by the Workers AI team and a new decision-model family, available on Workers AI today.","facts":"{\"title\":\"Workers AI - Introducing Clef: Cloudflare's first open-source decision models, now on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-01-clef-workers-ai/\",\"summary\":\"Meet @cf/cloudflare/clef and @cf/cloudflare/clef-flash , the first models trained by the Cloudflare Workers AI team, available on Workers AI today. Clef is a decision model, in the same family as Typesafe's Jev ↗︎ . Instead of generating text, it reads an input state and a set of typed questions, then returns a probability for every allowed answer. Your agent gets a structured decision it can act \",\"date\":\"2026-10-01\"}","detected_at":"2026-10-01","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"084d1b04226ee6afb0a1a0707f97561a42568a8ad4a5ed40857b2d8d6809ea26","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"5fef26e05d1e7cc36185f159a7eeb11309c50299b86c18e3ec269739e0edfe41","change_id":"5fef26e05d1e7cc36185f159a7eeb11309c50299b86c18e3ec269739e0edfe41","provider":"cloudflare","verdict":"major","rationale":"The entry announces first-class Agents SDK support for building agents on the Pi harness via a new PiHarness class, a new capability that is available now.","facts":"{\"title\":\"Agents, Workers - Run the Pi Durable harness on Cloudflare with the Agents SDK\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-02-pi-harness/\",\"summary\":\"The Agents SDK now provides first-class support for building agents using the Pi harness. You can build long-running agents using the combination of Pi 1.0 ↗︎ , Pi Durable ↗︎ , and the new PiHarness class that the Cloudflare Agents SDK provides, ensuring your agent's work is durably persisted, even if interrupted mid-turn. Built with Earendil ↗︎ , this integration is our first step toward first-cl\",\"date\":\"2026-10-02\"}","detected_at":"2026-10-02","judged_at":"2026-10-09T06:19:56.940Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-09T06:01:26.004Z"},
{"id":"b65e47dba44d235ff74f8013b0f9383ac0a19164886c053e2c1dea951091e167","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"1ecf58a556f51eaa9f1d485880e85090bba4249cc0d2b59576343d3b467fee7a","change_id":"1ecf58a556f51eaa9f1d485880e85090bba4249cc0d2b59576343d3b467fee7a","provider":"cloudflare","verdict":"minor","rationale":"Clef-omni joins the existing Clef family on Workers AI alongside a Clef-flash price cut, which is a single model added to an existing catalog plus a price change.","facts":"{\"title\":\"Workers AI - Clef-omni adds audio and video input, Clef-flash is now cheaper, and Clef is faster\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-09-clef-omni-workers-ai/\",\"summary\":\"@cf/cloudflare/clef-omni is now available on Workers AI. Clef-omni is a decision model that takes audio (WAV or MP3) and video (MP4 or WebM) input alongside text and images. It joins Clef and Clef-flash in the Clef family of open-weight decision models. We also cut the price of Clef-flash, so it now costs less than Jev, and made Clef faster. Clef-omni: one decision model for every modality Previou\",\"date\":\"2026-10-09\"}","detected_at":"2026-10-09","judged_at":"2026-10-10T00:18:23.973Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-10T00:01:52.525Z"},
{"id":"c2be154ea48f11b021c658782cc162fae1c776dbfa8aa14895cb6ce818ec9158","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"23924ba52e55594c2a5c58d514268d7634410fd87ea3ba5c8f32f15b3b29e7d8","change_id":"23924ba52e55594c2a5c58d514268d7634410fd87ea3ba5c8f32f15b3b29e7d8","provider":"cloudflare","verdict":"minor","rationale":"The entry adds Cloudflare skills to the existing Cloudflare API MCP server through an extension; it is not a new product, model family or major version, and the entry does not say it is generally available.","facts":"{\"title\":\"Agents - Cloudflare API MCP server serves Cloudflare skills\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-10-cloudflare-mcp-skills/\",\"summary\":\"The Cloudflare API MCP server now serves Cloudflare skills ↗︎ through the Skills over MCP extension ↗︎ . MCP clients that support the extension discover the skills with skills/list and read their files at skill:// / . To use them, add https://mcp.cloudflare.com/mcp to an MCP client that supports the extension ↗︎ .\",\"date\":\"2026-10-10\"}","detected_at":"2026-10-10","judged_at":"2026-10-10T06:18:27.928Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-10T06:01:59.571Z"},
{"id":"eb43f4dc5581c06ffc69211b536cf90828926d0607ea2b82c82104a73c87e43c","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"incident:openai:01M4H7FKXDA425TDNP2CRGEYA3","change_id":"402d855e77210c98e6a7bf031a8c66c52a51e48e2941678a40f9e6f53bb5aa00","provider":"openai","verdict":"not-material","rationale":"The incident \"Some ChatGPT content may be unavailable on Android\" names a consumer app, not the vendor API or a flagship model.","facts":"{\"title\":\"Some ChatGPT content may be unavailable on Android\",\"impact\":\"major\",\"started_at\":\"2026-10-09T21:00:39Z\",\"resolved_at\":null,\"hours\":3}","detected_at":"2026-10-10T00:00:50.871Z","judged_at":"2026-10-10T00:18:23.973Z","source_url":"https://status.openai.com/api/v2/incidents.json","fetched_at":"2026-10-10T00:00:50.871Z"},
{"id":"05548b5e1085962f8652a920414e15c0640ecf5625a574b8e1e94a2da83bc916","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:obtaining-public-statistics-with-disallowed-requests","change_id":"816935d136734072e8c387acf759f0ec95f4ead491bf08d456dcea2e7136bde3","provider":"openai","verdict":"not-material","rationale":"An OpenAI misalignment report observed during RL training on an internal research model; the disclosure record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Obtaining public statistics with disallowed requests\"}","detected_at":"2026-10-10T06:00:58.634Z","judged_at":"2026-10-10T06:18:27.928Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-10T06:00:58.634Z"},
{"id":"6c00c50f9b4436a841a6928fb31fdf52936bca264cd85fe609c9313fc2a93fa4","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:damaging-the-task-environment-to-trigger-a-reset","change_id":"ecd56e3753c5405978586f519cdedc736b7186b4ce3161fd94e5e5efb584980a","provider":"openai","verdict":"not-material","rationale":"An OpenAI misalignment report observed during RL training on an internal research model; the disclosure record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Damaging the task environment to trigger a reset\"}","detected_at":"2026-10-10T06:00:58.634Z","judged_at":"2026-10-10T06:18:27.928Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-10T06:00:58.634Z"},
{"id":"f9c69ad8c7d8ae8bef9f6cdec6598c722a35adc6ba9ed0e1d7d50e7dc7a6069b","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:sending-disallowed-web-requests-and-reaching-a-public-file-service","change_id":"35379a9be05385fffdff94292dbeb3d10e4c5a9e77cde59c850567cd6fca35e5","provider":"openai","verdict":"not-material","rationale":"An OpenAI misalignment report observed during RL training on an internal research model; the disclosure record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Sending disallowed web requests and reaching a public file service\"}","detected_at":"2026-10-10T06:00:58.634Z","judged_at":"2026-10-10T06:18:27.928Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-10T06:00:58.634Z"},
{"id":"9a2c9126ff95ed99fbdb2dc8a645f649349ace15acbbb8203324e71261018d27","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"d98f4d3470cf1b98e478713e973d2b8d0f6028365307b637282ae296b4748b3e","change_id":"d98f4d3470cf1b98e478713e973d2b8d0f6028365307b637282ae296b4748b3e","provider":"cloudflare","verdict":"minor","rationale":"A web_fetch tool added to the existing Agents SDK is a new option within a shipped SDK, not a new product, model family or major version.","facts":"{\"title\":\"Agents - New web_fetch tool in the Agents SDK for reading webpages and documents, with automatic markdown conversion\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-11-web-fetch/\",\"summary\":\"The Cloudflare Agents SDK now provides a web_fetch tool that reads webpages, PDFs, and supported Office documents, using Workers AI Markdown Conversion to turn them into text the model can read. The tool requests Markdown directly from sites that support Markdown for Agents and formats JSON for the model. When you give the web_fetch tool to your agent, the model provides a URL and receives back th\",\"date\":\"2026-10-12\"}","detected_at":"2026-10-12","judged_at":"2026-10-12T18:18:41.255Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-12T18:01:54.063Z"},
{"id":"b9548a6ad509a9b52c1888836e1385f211fa377171b05fd90d4e4e42d301785a","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"b667ff52653d8646bffd229269a68f8f9e1d4241159915e2ad573a5f2b81dfdc","change_id":"b667ff52653d8646bffd229269a68f8f9e1d4241159915e2ad573a5f2b81dfdc","provider":"cloudflare","verdict":"minor","rationale":"A web_search tool added to the existing Agents SDK is a new option within a shipped SDK; the entry does not announce a new product or major version.","facts":"{\"title\":\"Agents - New web_search tool in the Agents SDK, powered by the Cloudflare Web Search API\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-12-web-search/\",\"summary\":\"The Cloudflare Agents SDK now provides a web_search tool that searches the public web through the Cloudflare Web Search API , powered by ceramic.ai, Exa, and Linkup . When you give the web_search tool to your agent, the model provides a query and max results, and receives back a short numbered list of titles, URLs, and trimmed descriptions. When you also provide the web_fetch tool to your agent, t\",\"date\":\"2026-10-12\"}","detected_at":"2026-10-12","judged_at":"2026-10-12T18:18:41.255Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-12T18:01:54.063Z"},
{"id":"ceb68c44380fe35054840750fe97a583acec9cbb755306c7e9c5dd3ae77c77b0","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"772f7aa86307f51bc1537c74e570c5c088de4f34a30c9fd4cf2aa00681715dd2","change_id":"772f7aa86307f51bc1537c74e570c5c088de4f34a30c9fd4cf2aa00681715dd2","provider":"cloudflare","verdict":"minor","rationale":"OpenCode 2.0 support arrives through the experimental OpenCodeHarness, so it is not a generally available capability or a new product.","facts":"{\"title\":\"Agents, Workers - Run OpenCode 2.0 on Cloudflare with the Agents SDK\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-12-opencode-harness/\",\"summary\":\"The Agents SDK now supports OpenCode 2.0 ↗︎ through the experimental OpenCodeHarness . Run OpenCode inside a plain Durable Object, with sessions stored in SQLite and Lifecycle jobs that wake interrupted work after eviction. The agents/models/opencode provider connects OpenCode to Workers AI through the AI binding. OpenCode plugins extend the harness with custom tools, remote MCP servers, and skill\",\"date\":\"2026-10-12\"}","detected_at":"2026-10-12","judged_at":"2026-10-12T18:18:41.255Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-12T18:01:54.063Z"},
{"id":"a4ef64021e8a33c4b44a0729adb996f9d22a8bf4e874c5c1ae0d042d6c01e9c0","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"310db7442892e86069c21470cfe1fa1139dc6b931923ba9bfef7bf7e6b0fee60","change_id":"310db7442892e86069c21470cfe1fa1139dc6b931923ba9bfef7bf7e6b0fee60","provider":"cloudflare","verdict":"major","rationale":"The entry announces a new capability now provided in the Agents SDK: a persistent browser tool with tabs and logins that last between turns.","facts":"{\"title\":\"Agents - New persistent browser tool in the Agents SDK, with tabs and logins that last between turns\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-13-persistent-browser/\",\"summary\":\"The Cloudflare Agents SDK now provides a browserTool that gives your agent a Browser Run browser it can return to on later turns. Your agent can open a website, inspect a page, and come back to it with its tabs, cookies, and login session intact while the browser session remains alive. When you give browserTool to your agent, the model writes code that drives the browser through the Chrome DevTool\",\"date\":\"2026-10-13\"}","detected_at":"2026-10-13","judged_at":"2026-10-13T18:19:17.549Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-13T18:01:56.314Z"},
{"id":"d38d75a675933bd47e15c121e95023818c612e22b09044011c22486feb2ebf42","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"309efe355f9811eef7d952492f679a1204d5fb73c01e5eb8edecc6665e637bd1","change_id":"309efe355f9811eef7d952492f679a1204d5fb73c01e5eb8edecc6665e637bd1","provider":"cloudflare","verdict":"minor","rationale":"The entry extends the existing Cloudflare API MCP server to send Workers Issues as MCP Events; it is an added option on an existing product, not a new product, model family or major API version.","facts":"{\"title\":\"Agents, Workers - Cloudflare API MCP server sends Workers Issues as MCP Events\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-13-cloudflare-mcp-events/\",\"summary\":\"Workers Issues provides built-in error monitoring for Cloudflare Workers. It detects production failures and groups related failures into issues without an SDK or application wrapper. The Cloudflare API MCP server now sends these issues as MCP Events ↗︎ , the first event it exposes. MCP clients discover cloudflare.alert.workers_observability_real_time_issue with events/list and subscribe with even\",\"date\":\"2026-10-13\"}","detected_at":"2026-10-13","judged_at":"2026-10-13T12:19:04.433Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-13T12:02:03.855Z"},
{"id":"78cfb01873b9595357af99a611ad6655b393534d97f3f452a9e06826c64878bf","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"incident:anthropic:978mjkgw6mh7","change_id":"6f19bb39663f70e74ab73af2690c01fb4770ceffc2d55a01ae8ec4096b14ec8b","provider":"anthropic","verdict":"not-material","rationale":"The incident \"Elevated errors on platform.claude.com\" lists only the Claude Console (platform.claude.com) component, a console rather than the vendor API or a flagship model.","facts":"{\"title\":\"Elevated errors on platform.claude.com\",\"impact\":\"major\",\"started_at\":\"2026-10-07T13:25:04.155Z\",\"resolved_at\":\"2026-10-13T18:00:04.363Z\",\"hours\":148.6}","detected_at":"2026-10-13T18:00:50.777Z","judged_at":"2026-10-13T18:19:17.549Z","source_url":"https://status.claude.com/api/v2/incidents.json","fetched_at":"2026-10-13T18:00:50.777Z"},
{"id":"b35339548664f176f8d9981693ace2d50048eb5f5e8c77c00c804c1b028d48ef","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"dc79b351f8c0ae5a34ef332bbb0ccb2460c741b856d951945b09349538a6a377","change_id":"dc79b351f8c0ae5a34ef332bbb0ccb2460c741b856d951945b09349538a6a377","provider":"cloudflare","verdict":"minor","rationale":"The entry adds a single model, PhoneLLM Mini, to the existing Workers AI catalog, which the rubric rates minor.","facts":"{\"title\":\"Workers AI, Realtime - Pipecat PhoneLLM Mini now available on Workers AI\",\"products\":[\"Workers AI\",\"Realtime\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-15-phonellm-mini-workers-ai/\",\"summary\":\"@cf/daily-co/phonellm-mini-alpha-2 , a model from the Pipecat ↗︎ team at Daily, is now available on Workers AI . PhoneLLM is built for voice agents and other realtime AI applications, where a model has to reply quickly and call the right tool mid-conversation. In a voice agent, the language model sits between speech-to-text and text-to-speech, so every second it takes adds to the caller's wait. De\",\"date\":\"2026-10-15\"}","detected_at":"2026-10-15","judged_at":"2026-10-16T00:18:33.297Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-16T00:02:47.356Z"},
{"id":"2fa96bd879f10e71f976624da6d0f32911dc47e778ec7f307b2a2662de42d6e4","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"incident:openai:01M506BA4XRSWH7THR53HQGN3Z","change_id":"b86f2691b122059d946a5def6a702ce92d92fdf12ce229d91fb2505cac1d75f5","provider":"openai","verdict":"not-material","rationale":"The title \"Widespread feature degradation\" does not show the incident hit the vendor API or a flagship model, so under the rubric it is not material.","facts":"{\"title\":\"Widespread feature degradation\",\"impact\":\"major\",\"started_at\":\"2026-10-15T16:29:26Z\",\"resolved_at\":\"2026-10-15T18:56:00Z\",\"hours\":2.4}","detected_at":"2026-10-16T00:01:43.465Z","judged_at":"2026-10-16T00:18:33.297Z","source_url":"https://status.openai.com/api/v2/incidents.json","fetched_at":"2026-10-16T00:01:43.465Z"},
{"id":"24d7fdb67d2c1a0631333b6dd4f56f1207d88584cd6c2f68f1b5d68382678e8e","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:accessing-and-changing-internal-systems-during-dashboard-research","change_id":"e3a8d0035ce07923435a85c891eddb5a636b382e44c9a44e9aea408f9e615971","provider":"openai","verdict":"not-material","rationale":"The report \"Accessing and changing internal systems during dashboard research\" concerns an internal model in evaluation; the record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Accessing and changing internal systems during dashboard research\"}","detected_at":"2026-10-16T06:01:41.498Z","judged_at":"2026-10-16T06:20:23.803Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-16T06:01:41.498Z"},
{"id":"76fe1f350da367e56886a07cb7ba1721022021fc17e8b575d8d2628c6a19579f","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:retrieving-server-secrets-and-bypassing-authorization-on-a-government-statistics-website","change_id":"7de2116c609b5ca660e96a337f7a8b314a8753c343602f9c8fe4237b863dc08a","provider":"openai","verdict":"not-material","rationale":"The report \"Retrieving server secrets and bypassing authorization on a government statistics website\" concerns an internal model in RL training; the record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Retrieving server secrets and bypassing authorization on a government statistics website\"}","detected_at":"2026-10-16T06:01:41.498Z","judged_at":"2026-10-16T06:20:23.803Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-16T06:01:41.498Z"},
{"id":"b5d3cadd2a7b9eb0ecd7c281829614687c7432f726e7b5b8bc0f11849c5421bf","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:using-a-funded-test-cryptocurrency-wallet-without-authorization","change_id":"62664036518f7740ecdaaa88cef9c02392013bfc76a1ef19d48684812682060f","provider":"openai","verdict":"not-material","rationale":"The report \"Using a funded test cryptocurrency wallet without authorization\" concerns an internal research model on a coding task; the record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Using a funded test cryptocurrency wallet without authorization\"}","detected_at":"2026-10-16T06:01:41.498Z","judged_at":"2026-10-16T06:20:23.803Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-16T06:01:41.498Z"},
{"id":"d6a7060135948d68ed6426b4912670d7a279b787adbbe4f66f8c067e23d0cee4","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:circumventing-internet-restrictions-through-an-external-browser","change_id":"b6072a00d53ca3e2b29a2074b1b21fce518a1144c10adc9500607d19d63e3d3e","provider":"openai","verdict":"not-material","rationale":"The report \"Circumventing internet restrictions through an external browser\" concerns an internal research model in RL training; the record carries no high-severity label and names no vendor API or flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Circumventing internet restrictions through an external browser\"}","detected_at":"2026-10-16T06:01:41.498Z","judged_at":"2026-10-16T06:20:23.803Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-16T06:01:41.498Z"},
{"id":"349c2ffe8d866c3069112699178eb9442aa9455a4f794994d72bcb80d9346e4e","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"47d2ec086bfe32ae1d8fa8d18c064a856556e0db4635581075e2122882f7d7ed","change_id":"47d2ec086bfe32ae1d8fa8d18c064a856556e0db4635581075e2122882f7d7ed","provider":"cloudflare","verdict":"minor","rationale":"Step-up authorization on the existing Cloudflare API MCP server is a small option on an existing product, not a new product or capability launch.","facts":"{\"title\":\"Agents - Cloudflare API MCP server now supports step-up authorization\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-19-cloudflare-mcp-step-up/\",\"summary\":\"The Cloudflare API MCP server now supports step-up authorization ↗︎ . Grant only the required scopes when you connect, then authorize more on demand as your agent performs a task. When a tool call needs a scope you have not granted, the server returns an insufficient_scope challenge ↗︎ naming it, and your MCP client asks you to approve it. Scopes you already granted are kept. To start with the lea\",\"date\":\"2026-10-19\"}","detected_at":"2026-10-19","judged_at":"2026-10-19T18:18:37.970Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-19T18:01:36.014Z"},
{"id":"469b4efa25a0162e8146947f164b160ca13a30d8408cbae27cfe41b6cca0e489","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"3105c74f523c5756af96b16c1aa42b7f572d8bf843391124c50a0ef633214f81","change_id":"3105c74f523c5756af96b16c1aa42b7f572d8bf843391124c50a0ef633214f81","provider":"cloudflare","verdict":"minor","rationale":"The entry \"Workers AI - DeepSeek V4.1 Flash 1017 now available on Workers AI\" adds a single model update to the existing Workers AI catalog, which the rubric rates minor.","facts":"{\"title\":\"Workers AI - DeepSeek V4.1 Flash 1017 now available on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-19-deepseek-v4.1-flash-workers-ai/\",\"summary\":\"@cf/deepseek-ai/deepseek-v4.1-flash-1017 is now available on Workers AI. DeepSeek V4.1 Flash is a multimodal Mixture-of-Experts model with 552B backbone parameters. It activates 8B parameters per token during prefill and 16B during decode, which keeps input-heavy agentic workloads fast and low-cost. The 1017 release is an update to DeepSeek V4.1 Flash with improvements across reasoning, coding, ag\",\"date\":\"2026-10-19\"}","detected_at":"2026-10-19","judged_at":"2026-10-20T06:20:20.353Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-20T06:01:30.149Z"},
{"id":"a3cf554efea56f6ee63ec5ebd270bbe1c97a3d438e12696de72513808f59564b","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"29558137cf3a7985f00a12260738db8c36ace7ba01cf21d87760b276de05ddf4","change_id":"29558137cf3a7985f00a12260738db8c36ace7ba01cf21d87760b276de05ddf4","provider":"cloudflare","verdict":"major","rationale":"The entry introduces Clairic, described as the first extraction model, a new kind of model rather than another model added to an existing catalog, so it reads as a new model family.","facts":"{\"title\":\"Workers AI - Introducing Clairic: an extraction model for typed values, now on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-19-clairic-workers-ai/\",\"summary\":\"@cf/cloudflare/clairic is now available on Workers AI as a preview. Clairic is the first extraction model : a System One model that answers with values instead of text. We intend to release the model with open weights soon. Decision models such as Clef choose among options you define up front. Many structured-output tasks do not have answers you can list in advance: an invoice total, a delivery da\",\"date\":\"2026-10-19\"}","detected_at":"2026-10-19","judged_at":"2026-10-19T18:18:37.970Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-19T18:01:36.014Z"},
{"id":"ea78e2074a672d23fd24eef25b3eaa62c436954f1f2f0a3c303802de1bd48f45","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"8bfb73b614d69bb8acc08e041a5a3d07065e690e12fb3026c4695c5af1f82d0e","change_id":"8bfb73b614d69bb8acc08e041a5a3d07065e690e12fb3026c4695c5af1f82d0e","provider":"cloudflare","verdict":"minor","rationale":"The entry moves Think and Agent Skills onto the @cloudflare/computer workspace and deprecates @cloudflare/shell; that is a change to an existing product plus a deprecation, not a new product or a generally available new capability.","facts":"{\"title\":\"Agents, Workers - Think agents now use a @cloudflare/computer workspace\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-20-think-computer-workspace/\",\"summary\":\"Think and Agent Skills now use the workspace from @cloudflare/computer ↗︎ as well as the filesystem tools. A JavaScript execution environment can be configured by adding the JavaScriptWorkerBackend . @cloudflare/shell is deprecated, and so is Think's createExecuteTool . These will no longer be developed nor supported from today onwards. import { Think } from \\\"@cloudflare/think\\\" ; import { Workspac\",\"date\":\"2026-10-20\"}","detected_at":"2026-10-20","judged_at":"2026-10-20T18:18:30.666Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-20T18:01:34.810Z"},
{"id":"8e2250c9f0c5bb4d5c2eb4aab18be11f678eecb0b07aeb26673f67bb5a55ec96","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"6fea97ff164ddce724bb54466ef9ab8467461d4b582ef3c338c2c50f1a003d70","change_id":"6fea97ff164ddce724bb54466ef9ab8467461d4b582ef3c338c2c50f1a003d70","provider":"cloudflare","verdict":"minor","rationale":"createAI consolidates what workers-ai-provider and ai-gateway-provider already did into one Agents SDK function and deprecates those two, a repackaging plus deprecation rather than a new product or major version.","facts":"{\"title\":\"Agents, Workers AI, AI Gateway - createAI replaces workers-ai-provider and ai-gateway-provider\",\"products\":[\"Agents\",\"Workers AI\",\"AI Gateway\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-21-createai/\",\"summary\":\"The Agents SDK has a new createAI function for the AI SDK and Pi AI . It unifies what workers-ai-provider and ai-gateway-provider did: Workers AI models, AI Gateway catalog models, dynamic routes, and vendor models, through the AI binding or an account id and API token. workers-ai-provider and ai-gateway-provider are now feature frozen and marked as deprecated. Please migrate at your convenience. \",\"date\":\"2026-10-21\"}","detected_at":"2026-10-21","judged_at":"2026-10-21T12:19:21.482Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-21T12:01:30.983Z"},
{"id":"433b71a04261d0f5e38f30e71128cee0c2d35153c83fa360a2763ae50c5389a8","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:suppressing-misalignment-monitoring-alerts-and-disabling-misalignment-monitoring","change_id":"4780e9895b545e9f1d688b3c0b840bb728cf286b9976db3bfac336fa1b45ec54","provider":"openai","verdict":"not-material","rationale":"The report \"Suppressing misalignment monitoring alerts and disabling misalignment monitoring\" does not label the event high severity and does not describe a hit to a vendor API; it is a disclosure from internal use, so it does not meet the material bar.","facts":"{\"kind\":\"report\",\"title\":\"Suppressing misalignment monitoring alerts and disabling misalignment monitoring\"}","detected_at":"2026-10-21T00:00:43.791Z","judged_at":"2026-10-21T00:18:01.656Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-10-21T00:00:43.791Z"},
{"id":"1144754ebe5ace5da28ccb6c29a2429fcd438778de7726901d8e004fc60ea848","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"ad037315fb7321e80bf221a723a56b71a49f0d937f558d41b5c8741e619629d4","change_id":"ad037315fb7321e80bf221a723a56b71a49f0d937f558d41b5c8741e619629d4","provider":"cloudflare","verdict":"minor","rationale":"The entry retires the /sse endpoint on Cloudflare MCP servers in favour of /mcp, which is a deprecation rather than a new product or capability.","facts":"{\"title\":\"Agents - Cloudflare MCP servers no longer serve /sse\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-22-cloudflare-mcp-servers-retire-sse/\",\"summary\":\"Cloudflare's product-specific MCP servers now serve only /mcp . A request to /sse returns 410 Gone with the server's /mcp URL, which most MCP clients show in their connection error. If your MCP client is configured with an /sse URL, change it to /mcp , for example https://docs.mcp.cloudflare.com/mcp . If you signed in through /sse , your client asks you to sign in again. Connections that already u\",\"date\":\"2026-10-22\"}","detected_at":"2026-10-22","judged_at":"2026-10-22T18:18:45.776Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-22T18:01:41.112Z"},
{"id":"98e8960a80ac7dc28f6e5278cfbb6d421b4906763467931c8418709cc4edd0e2","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"8d56cff60c2ebcfd02a36a8fae25340184a55cf2745f6f74e32d746561ceb55a","change_id":"8d56cff60c2ebcfd02a36a8fae25340184a55cf2745f6f74e32d746561ceb55a","provider":"cloudflare","verdict":"minor","rationale":"The entry updates the Agents SDK to support a newer Pi release with reliability improvements, an incremental change rather than a new product or a new major SDK version.","facts":"{\"title\":\"Agents - Agents SDK now supports the latest Pi 1.4 release\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-22-pi-harness-pi-1-4/\",\"summary\":\"Agents SDK v0.30.0 runs PiHarness on Pi 1.4 ↗︎ , and makes it recover more reliably from evictions, crashes, and storage errors. Hibernation while Pi's tasks sleep. PiHarness passes Pi's new onWake callback to your harness factory. While a task sleeps, such as during a model retry backoff, the Durable Object can hibernate until the task is due instead of staying in memory. Parked tasks. Pi 1.4 par\",\"date\":\"2026-10-22\"}","detected_at":"2026-10-22","judged_at":"2026-10-22T18:18:45.776Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-22T18:01:41.112Z"},
{"id":"28f25ac7e6dd0b5ed2051f4ee5773c56ce4280cc6faaf0b09ca42b47bf99aba5","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"fb0db9fc623d2e5d5178defa983e2e11f3141f046f5b0aac4e89e45eeb97bc88","change_id":"fb0db9fc623d2e5d5178defa983e2e11f3141f046f5b0aac4e89e45eeb97bc88","provider":"cloudflare","verdict":"minor","rationale":"The entry adds a resume-after-interruption option (onInterrupt: \"replay\") and approval pauses to the existing @cloudflare/computer package; that is a small option on an existing product, not a new product, model family or major SDK version.","facts":"{\"title\":\"Agents, Workers - Cloudflare Computer now helps agents recover from interruptions and wait for approval\",\"products\":[\"Agents\",\"Workers\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-26-computer-durable-javascript/\",\"summary\":\"@cloudflare/computer ↗︎ gives your agent a workspace to run code in. From v0.7.0, that code can resume after a deploy or crash, and pause on an external action, for example to approve a tool call. Resume interrupted runs Set onInterrupt: \\\"replay\\\" on WorkerJavaScriptBackend , or on a single run. import { DurableObject } from \\\"cloudflare:workers\\\" ; import { Workspace } from \\\"@cloudflare/computer\\\" ; \",\"date\":\"2026-10-26\"}","detected_at":"2026-10-26","judged_at":"2026-10-26T18:18:56.526Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-26T18:02:21.455Z"},
{"id":"f76f548f32bbf3173b93fde4760dddeb62c8473c1a4f05ef0184db08e3db74c5","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"9677297451924a3aac3c4e4613fdd74ae40923906cff0d2ad855ae30e808438f","change_id":"9677297451924a3aac3c4e4613fdd74ae40923906cff0d2ad855ae30e808438f","provider":"cloudflare","verdict":"minor","rationale":"The entry adds internal hostnames as an option for the existing persistent browser in the Cloudflare Agents SDK, a small capability on an existing product rather than a new product or major SDK version.","facts":"{\"title\":\"Agents, Browser Run - Agents SDK: Preview dev servers in the browser without public URLs\",\"products\":[\"Agents\",\"Browser Run\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-28-browser-internal-hostnames/\",\"summary\":\"Agents that build websites need to check their work in a real browser. Internal hostnames let your agent's browser preview a dev server running in a container without exposing that server to the public Internet. With internal hostnames, the persistent browser in the Cloudflare Agents SDK can open sites that only your agent serves. Give the browser hostnames such as preview.internal , and its reque\",\"date\":\"2026-10-28\"}","detected_at":"2026-10-28","judged_at":"2026-10-28T18:19:05.393Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-28T18:02:58.388Z"},
{"id":"5e54776b03a67cfbc34145c1d96113a62c2eeee5fd1f7ea4bc986357a250fc19","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"53205cef201b977d7de16bae5940ab1350c10ec966c00b3b2aa06ac26ef405f0","change_id":"53205cef201b977d7de16bae5940ab1350c10ec966c00b3b2aa06ac26ef405f0","provider":"cloudflare","verdict":"minor","rationale":"The entry changes how Agent Skills are loaded in the existing Agents SDK, letting skills be files in the Workspace; it is an option on an existing feature, not a new product or a new major SDK version.","facts":"{\"title\":\"Agents - Agents SDK: Agent Skills are files in your agent's Workspace\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-10-29-agent-skills-from-files/\",\"summary\":\"Agent Skills in the Cloudflare Agents SDK are now files. Write a folder with a SKILL.md into your agent's Computer ↗︎ Workspace, under /workspace/.agents/skills/ , and the agent can use it on its next turn. You do not need a build step or a deploy, and the agent can write its own skills. src/server.js js import { Think, skills } from \\\"@cloudflare/think\\\" ; export class MyAgent extends Think { getSk\",\"date\":\"2026-10-29\"}","detected_at":"2026-10-29","judged_at":"2026-10-29T18:20:44.953Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-10-29T18:02:30.416Z"},
{"id":"f71d3f10a0b266367e41434eb90050bceabb2faf4bf659f7019b2bb8bc0b6b13","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"4968cd57fc554c5d8caeb70d9896638b75855caed9924191f60845b000044d82","change_id":"4968cd57fc554c5d8caeb70d9896638b75855caed9924191f60845b000044d82","provider":"cloudflare","verdict":"minor","rationale":"The entry is a performance upgrade to the existing Agents SDK harness (fewer storage writes, lower memory use), not a new product, a newly generally available capability, or a new model family.","facts":"{\"title\":\"Agents - Agents SDK now supports Pi 1.5 with fewer storage writes and lower memory use\",\"products\":[\"Agents\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-11-03-pi-harness-pi-1-5/\",\"summary\":\"Agents SDK v0.33.0 upgrades PiHarness to Pi 1.5 ↗︎ . The release reduces storage transactions while streaming, improves throughput across many sessions, and reduces memory use when streaming large tool calls. Performance improvements Far fewer storage writes while streaming. Pi stores streamed answers and tool output in batches, within a second of each update, instead of on every update. Over SQLi\",\"date\":\"2026-11-03\"}","detected_at":"2026-11-03","judged_at":"2026-11-03T18:19:23.138Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-11-03T18:02:18.915Z"},
{"id":"017394087587daec5808ef442024898b84ff81c294cc184aec1cc3a62d1f9b13","thesis_id":"cloudflare","claim_id":"cloudflare-ai-launches","subject":"b937e0b1ca3ca048bf77b163b62aa3c7490c2164674fbec954d9f54f3f3dbae7","change_id":"b937e0b1ca3ca048bf77b163b62aa3c7490c2164674fbec954d9f54f3f3dbae7","provider":"cloudflare","verdict":"minor","rationale":"The entry title is \"Workers AI - Planned model deprecations on Workers AI\" and the summary describes removing deprecated models from the catalog before November 15, 2026 — a deprecation, which the rubric rates minor.","facts":"{\"title\":\"Workers AI - Planned model deprecations on Workers AI\",\"products\":[\"Workers AI\"],\"link\":\"https://developers.cloudflare.com/changelog/post/2026-11-05-planned-model-deprecations/\",\"summary\":\"We are refreshing the Workers AI model catalog to make room for new releases and updated models. Update your apps to remove references to the models listed below before November 15, 2026. We recommend moving to newer models for the best token efficiency and throughput. For pricing, refer to the Workers AI pricing page . Recommended LLM replacements For frontier model needs: @cf/deepseek-ai/deepsee\",\"date\":\"2026-11-05\"}","detected_at":"2026-11-05","judged_at":"2026-11-06T06:21:44.444Z","source_url":"https://developers.cloudflare.com/changelog/rss/index.xml","fetched_at":"2026-11-06T06:02:28.653Z"},
{"id":"a05dda4a258300a69eb1ab417bc8bb8596452861ebb166b84c8981689ad61ad8","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:covert-underperformance-on-safety-research-tasks","change_id":"185926de70e97ef29c4624ee10fad72220a31563450d1ae9900d16f171a65e2e","provider":"openai","verdict":"not-material","rationale":"The disclosure says the behavior was observed during \"Internal deployment and evals,\" not in the vendor API or a flagship model's live traffic, and the record carries no severity label, so it does not meet the rubric's bar for material.","facts":"{\"kind\":\"report\",\"title\":\"Covert underperformance on safety research tasks\"}","detected_at":"2026-11-05T00:01:44.170Z","judged_at":"2026-11-05T04:08:34.187Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-11-05T00:01:44.170Z"},
{"id":"420a878d46c22e6cde51259f462e7dfb2518726698df178588cfd96a9cd76805","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:bypassing-network-restrictions-during-a-coding-evaluation","change_id":"16c9e6ddfbea007538664edee76d10083a681882b2aebc41cbd136781d45eab6","provider":"openai","verdict":"not-material","rationale":"The report concerns an Internal research model observed during Evals, not the vendor API or a flagship model, and carries no high-severity label, so it misses the material bar.","facts":"{\"kind\":\"report\",\"title\":\"Bypassing network restrictions during a coding evaluation\"}","detected_at":"2026-11-06T06:01:31.340Z","judged_at":"2026-11-06T06:21:44.444Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-11-06T06:01:31.340Z"},
{"id":"7668db57715013e2ec6996318b6412dd9b884d5c5203036ec8f93edaadeb328e","thesis_id":"anthropic","claim_id":"anthropic-safety-asset","subject":"disclosure:openai:report:trying-to-prompt-inject-the-grader-in-filenames","change_id":"035003a4ebf6dc249dee18ac2fd9dfcbd4fc4a3a8ca53d65c5b2751d6ca50611","provider":"openai","verdict":"not-material","rationale":"Like the other candidate, this report is about an Internal research model during Evals with no high-severity label and no indication it hit the vendor API or a flagship model.","facts":"{\"kind\":\"report\",\"title\":\"Trying to prompt-inject the grader in filenames\"}","detected_at":"2026-11-06T06:01:31.340Z","judged_at":"2026-11-06T06:21:44.444Z","source_url":"https://alignment.openai.com/misalignment-reports/","fetched_at":"2026-11-06T06:01:31.340Z"}
]