The Wire · Showcase
CODEX SHIPS CLI THREAD ARCHIVE, FIXES BEDROCK SERVICE TIER LEAKS
By RepoJournal · Filed · About OpenAI
Codex now lets you archive threads from the command line, while a critical fix prevents the system from advertising unsupported Bedrock service tiers.
The top story is straightforward: `codex archive` and `codex unarchive` commands are live [1], finally closing the gap between app-server RPC capabilities and what the CLI exposed. These commands resolve UUIDs or thread names and support scoped remote flags for multi-endpoint deployments. In parallel, a critical fix landed that strips non-default service tier metadata from Bedrock model catalogs [2], preventing Codex from advertising GPT models with tiers Bedrock doesn't actually support. This is a silent failure prevention: end users would have hit errors trying to use those tiers. On the multi-agent front, the v2 assignment tool got renamed from `followup_task` to `assign_task` [3] with backward compatibility preserved in rollout traces, keeping legacy classification intact while modernizing the API surface.
Action items
- → Deploy Codex CLI thread archive fix to production openai/codex [plan]
- → Verify Bedrock model catalog stripping is active in your environment openai/codex [monitor]
- → Update any internal docs referencing followup_task to assign_task openai/codex [plan]
References
- [1] Add thread archive CLI commands ↗ openai/codex
- [2] fix: Limit Bedrock GPT models to default service tier ↗ openai/codex
- [3] Rename multi-agent v2 assignment tool ↗ openai/codex
FAQ
- What changed in OpenAI on May 31, 2026?
- Codex now lets you archive threads from the command line, while a critical fix prevents the system from advertising unsupported Bedrock service tiers.
- What should OpenAI teams do about it?
- Deploy Codex CLI thread archive fix to production • Verify Bedrock model catalog stripping is active in your environment • Update any internal docs referencing followup_task to assign_task
- Which OpenAI repositories shipped on May 31, 2026?
- openai/codex