Daily digest · 2026-10-26 (UTC)
What broke for AI agents on 26 Oct 2026
6 incidents on 4 status pages.
Final. Built 2026-10-31 11:03 UTC, 7 hours after the day ended; this page will not change.
What broke
- npm: “View package and auth outage on npmjs.com”major impact · opened 05:54 UTC · resolved 06:30 UTC (36 minutes) · www.npmjs.com websitestatus page ↗provider-incidents v63
- Cloudflare: “Cloudflare Workers build delays”minor impact · opened 15:12 UTC · resolved 16:03 UTC (51 minutes) · Workers Buildsstatus page ↗provider-incidents v65
- Cloudflare: “Delayed, missed, and duplicated Workers Cron Triggers”minor impact · opened 21:48 UTC · resolved Oct 27 01:15 UTC (3 h 26 min) · Workersstatus page ↗provider-incidents v66
- Cloudflare: “Network Performance Issues in Chicago (ORD)”minor impact · opened 09:21 UTC · resolved 09:03 UTC (42 minutes)status page ↗provider-incidents v64
- GitHub: “Incident with Actions, Issues, API Requests and Pull Requests”minor impact · opened 18:48 UTC · resolved 18:57 UTC (9 minutes) · API Requests | Actions | Issues | Pull Requestsstatus page ↗provider-incidents v65
- OpenAI: “Elevated error rates for Ads POST requests”the feed gives no impact level · the feed gives no start time · resolved 18:30 UTC · Ads APIstatus page ↗provider-incidents v65
Shipped
No new major version of the SDKs and tools we follow.
22 other releases on the day
- @anthropic-ai/claude-agent-sdk 0.3.311 (npm)released 19:49 UTC · previous 0.3.310agent-sdk-releases v83
- ai 7.0.154 (npm)released 18:56 UTC · previous 7.0.153agent-sdk-releases v82
- anthropics/claude-code v2.1.311published 21:51 UTC · previous v2.1.310release notes ↗agent-tool-releases v83
- langchain-ai/langchain langchain-openai==1.7.3published 23:08 UTC · previous langchain-openai==1.7.2release notes ↗agent-tool-releases v83
- langchain-ai/langgraph 1.2.18published 20:04 UTC · previous 1.2.17release notes ↗agent-tool-releases v83
- langchain-ai/langgraph checkpoint==4.3.1published 19:57 UTC · previous checkpoint==4.3.0release notes ↗agent-tool-releases v83
- langchain-ai/langgraph cli==0.4.39published 14:16 UTC · previous cli==0.4.38release notes ↗agent-tool-releases v82
- langchain-ai/langgraph cli==0.4.40published 16:45 UTC · previous cli==0.4.39release notes ↗agent-tool-releases v82
- langchain-ai/langgraph sdk==0.4.10published 19:59 UTC · previous sdk==0.4.9release notes ↗agent-tool-releases v83
- openai 3.38.0 (pypi)released 16:36 UTC · previous 3.37.0agent-sdk-releases v82
- openai 3.39.0 (pypi)released 19:19 UTC · previous 3.38.0agent-sdk-releases v83
- openai 7.44.0 (npm)released 15:39 UTC · previous 7.43.0agent-sdk-releases v82
- openai 7.45.0 (npm)released 19:27 UTC · previous 7.44.0agent-sdk-releases v83
- openai/openai-node v7.44.0published 15:31 UTC · previous v7.43.0release notes ↗agent-tool-releases v82
- openai/openai-node v7.45.0published 19:14 UTC · previous v7.44.0release notes ↗agent-tool-releases v83
- openai/openai-python v3.38.0published 16:35 UTC · previous v3.37.0release notes ↗agent-tool-releases v82
- openai/openai-python v3.39.0published 19:14 UTC · previous v3.38.0release notes ↗agent-tool-releases v83
- vercel/ai @ai-sdk/[email protected]published 19:01 UTC · previous @ai-sdk/[email protected]release notes ↗agent-tool-releases v82
- vercel/ai @ai-sdk/[email protected]published 19:01 UTC · previous @ai-sdk/[email protected]release notes ↗agent-tool-releases v82
- vercel/ai [email protected]published 18:54 UTC · previous [email protected]release notes ↗agent-tool-releases v82
- vercel/ai [email protected]published 18:55 UTC · previous [email protected]release notes ↗agent-tool-releases v82
- vercel/ai [email protected]published 18:57 UTC · previous [email protected]release notes ↗agent-tool-releases v82
Measured on the day
- 184 round-trips to 23 endpoints agents depend on, 0 failedmodel APIs, registries, chains, the x402 facilitator, status pages and two baselines · measured from macbook-kragent-api-observatory v108
- 451 of 500 remote MCP endpoints probed answered a handshake (90.2%), on 483 hostsa different slice of the official registry each day, so the share is not compared across days · one attempt per endpoint, an auth challenge counts as an answer · measured from macbook-krmcp-server-liveness v25
How this is made
- Every line comes from one of WITAN's public datasets and links the version its records landed in: provider-incidents, agent-api-observatory, mcp-server-liveness, agent-sdk-releases, agent-tool-releases. They are free to read.
- Incidents are what the providers' own status pages published (Anthropic, OpenAI, GitHub, Cloudflare, npm, Vercel), read every 6 hours; scheduled maintenance is left out. Follow a line's status page link for the provider's text.
- Round-trips and MCP handshakes are WITAN's own measurements from one vantage point, named on each line. A failure there can be the network between us and the service; it is not a verdict on the service. A latency line needs the day's median at least 2× the previous 7 days' and 250 ms more.
- Releases are those of the SDKs and agent tools the datasets follow; pre-releases are left out.
- Days are UTC. A day is rebuilt every hour while it runs and frozen 7 hours after it ends. Model prices are not here: that dataset's collector is off. WITAN is a testnet preview run by one developer.
Earlier days
2026-11-012026-10-312026-10-302026-10-292026-10-282026-10-272026-10-262026-10-252026-10-24