Daily digest · 2026-11-16 (UTC)
What broke for AI agents on 16 Nov 2026
7 incidents on 4 status pages.
Final. Built 2026-11-17 07:30 UTC, 7 hours after the day ended; this page will not change.
What broke
- Anthropic: “Claude unable to join meetings”major impact · opened 15:10 UTC · identified as of Nov 17 00:00 UTC · claude.aistatus page ↗provider-incidents v132
- Anthropic: “Elevated errors for Claude Sonnet 5.5 and Claude Fable 5.2”major impact · opened 07:49 UTC · resolved 09:04 UTC (75 minutes) · Claude API (api.anthropic.com) | Claude Code | Claude Cowork | claude.aistatus page ↗provider-incidents v129
- Anthropic: “Some users unable to sign up with Apple on the Claude iOS app”major impact · opened Nov 14 07:11 UTC · resolved 19:52 UTC (3 days) · claude.aistatus page ↗provider-incidents v131
- Cloudflare: “Errors creating predictions for some models”minor impact · opened 14:30 UTC · resolved 16:19 UTC (109 minutes) · Replicatestatus page ↗provider-incidents v130
- Cloudflare: “Network Performance Issues in Montreal, Canada”minor impact · opened 21:36 UTC · resolved 20:40 UTC (30 minutes) · Montréal, Canada - (YUL)status page ↗provider-incidents v131
- GitHub: “Incident with Pull Requests”minor impact · opened 11:59 UTC · resolved 16:26 UTC (4 h 27 min) · Actions | Pull Requestsstatus page ↗provider-incidents v130
- OpenAI: “Delayed Costs Data in the Compliance API”the feed gives no impact level · first seen by our reader Nov 13 18:20 UTC (the feed gives no start time) · resolved 00:17 UTC · Compliance APIstatus page ↗provider-incidents v128
Shipped
No new major version of the SDKs and tools we follow.
13 other releases on the day
- @anthropic-ai/claude-agent-sdk 0.3.326 (npm)released 18:44 UTC · previous 0.3.325agent-sdk-releases v146
- @anthropic-ai/sdk 1.9.0 (npm)released 16:53 UTC · previous 1.8.0agent-sdk-releases v145
- @google/genai 3.4.0 (npm)released 21:04 UTC · previous 3.3.0agent-sdk-releases v146
- anthropic 2.9.0 (pypi)released 16:47 UTC · previous 2.8.0agent-sdk-releases v145
- anthropics/anthropic-sdk-python v2.9.0published 16:47 UTC · previous v2.8.0release notes ↗agent-tool-releases v143
- anthropics/anthropic-sdk-typescript sdk-v1.9.0published 16:47 UTC · previous sdk-v1.8.0release notes ↗agent-tool-releases v143
- anthropics/claude-agent-sdk-python v0.2.183published 23:06 UTC · previous v0.2.182release notes ↗agent-tool-releases v144
- anthropics/claude-code v2.1.326published 22:11 UTC · previous v2.1.325release notes ↗agent-tool-releases v144
- BerriAI/litellm v1.108.10published 21:57 UTC · previous v1.108.9release notes ↗agent-tool-releases v144
- claude-agent-sdk 0.2.183 (pypi)released 23:03 UTC · previous 0.2.182agent-sdk-releases v146
- google-genai 3.4.0 (pypi)released 21:45 UTC · previous 3.3.0agent-sdk-releases v146
- openai 3.52.0 (pypi)released 22:41 UTC · previous 3.51.0agent-sdk-releases v146
- openai/openai-python v3.52.0published 22:40 UTC · previous v3.51.0release notes ↗agent-tool-releases v144
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 v192
- 458 of 800 remote MCP endpoints probed answered a handshake (57.3%), on 542 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 v46
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
2026-11-212026-11-202026-11-192026-11-182026-11-172026-11-162026-11-152026-11-142026-11-132026-11-122026-11-112026-11-102026-11-092026-11-082026-11-072026-11-062026-11-052026-11-042026-11-032026-11-022026-11-012026-10-312026-10-302026-10-292026-10-282026-10-272026-10-262026-10-252026-10-24