GitHub Recovers, Codex Degraded
No internet-wide outage is underway this morning (as of about 09:00 UTC on October 13). AWS, Google Cloud, Azure, Akamai, Discord, Slack, Atlassian and Stripe all report normal operation. The notable news is a busy afternoon on October 12 that came after yesterday's report was filed, plus one open incident at OpenAI.
**GitHub: multi-hour disruption to Actions, Issues and Pull Requests (resolved).** Starting around 12:21 UTC on October 12, GitHub reported degraded performance across Actions, Issues and Pull Requests. Users had trouble loading the Actions tab, queuing workflow runs, creating or updating issues and requesting pull request reviews. Pull Requests recovered by 13:02 UTC, and Actions and Issues were mitigated by 14:36 UTC. Web and email notifications ran late for several more hours while a backlog cleared, and GitHub marked the incident resolved at 20:16 UTC. GitHub says a root cause analysis will follow. Source: https://www.githubstatus.com
**Vercel: build and deployment failures (resolved).** Vercel reported increased build and deployment failures from 13:35 to 20:11 UTC on October 12. Part of that came from the GitHub incident: GitHub-connected builds were affected upstream. Failure rates had dropped by 16:10 UTC and the deployment event backlog was processed by 19:56 UTC. Source: https://www.vercel-status.com
**OpenAI: degraded performance in Codex and ChatGPT Work mode (ongoing).** OpenAI is investigating degraded performance in GPT-6.1 Sol Codex and Work mode, affecting ChatGPT and Codex. The incident was opened early on October 13 UTC. OpenAI has posted only an initial "investigating" notice so far, with no detail on scope or cause. Source: https://status.openai.com
**Cursor: overnight errors (mostly recovered).** From 00:22 UTC on October 13, Cursor saw errors across its IDE, Cloud Agents, Automations and related services. A fix went in at 00:53 UTC, and by 03:07 UTC everything had recovered except Cloud Agents, which some users may still find slow to start. Source: https://status.cursor.com
**Cloudflare: several regional, minor incidents.** On October 12 Cloudflare logged network performance issues in Seattle (12:54 to 21:06 UTC), a broader CDN/cache performance incident (16:22 to 23:49 UTC), a short Workers API problem and a Workers Cron Trigger activation issue (19:09 to 22:20 UTC). All are resolved. Still open: elevated errors in Los Angeles (since 08:52 UTC October 12, fix in progress), congestion affecting its China Network, and a performance issue in Ashburn, Virginia, where a vendor's capacity restoration caused impact from 06:33 to 06:41 UTC today. Cloudflare says capacity there is restored and it is monitoring. Source: https://www.cloudflarestatus.com
**Other recently resolved incidents (past 72 hours):** DigitalOcean had intermittent external DNS lookup failures for Managed Kubernetes clusters in several regions on October 11 (22:01 to 23:29 UTC). Zoom had small, separate issues with Chat file transfers (October 10), Scheduler (October 11) and AI Companion for US users (October 12). Akamai closed its long-running APAC origin-connectivity incident, which it linked to subsea cable maintenance, late on October 12.
**Minor, ongoing:** Anthropic is still monitoring elevated errors on platform.claude.com (open since October 7). Twilio has carrier-specific SMS/MMS and voice delays on particular routes, including Thailand and Malaysia voice and Australia MMS. These affect individual routes, not Twilio as a whole.
We found no reports of major BGP incidents, DNS root problems or new undersea cable failures. Search coverage of the past 24 hours is still thin, so smaller regional events may not show up yet.