Summary

After updating OpenClaw, embedded agent runs fail with a 400 from the Azure OpenAI Responses provider:

Item '<id>' of type 'reasoning' was provided without its required following item.

Exact Error

00:43:37 [agent/embedded] embedded run agent end: runId=c79f048b-b5aa-4f4b-8a7d-f49d46b7f364 isError=true model=gpt-5.3-codex provider=azure-openai-responses error=400 Item 'rs_0b7b72782f7ad3800069c43058ebcc8194ab5903cdaab82f9b' of type 'reasoning' was provided without its required following item. rawError=400 Item 'rs_0b7b72782f7ad3800069c43058ebcc8194ab5903cdaab82f9b' of type 'reasoning' was provided without its required following item.

Environment

• OpenClaw: 2026.3.23-2 (7ffe7e4)
• Provider: azure-openai-responses
• Model: gpt-5.3-codex

Repro

1. Update OpenClaw to 2026.3.23-2.
2. Run embedded agent with Azure OpenAI Responses + gpt-5.3-codex.
3. After running, request fails with the 400 above.

Expected

Request should complete normally without transcript/reasoning item validation errors.

Notes

This appears related to recent reasoning/transcript-history issues (possibly similar to #53506 / #49167), but this report includes a fresh repro and exact payload.