LLM model/API behavior drift
Claude API thinking summary extended thinking redacted_thinking thinking block text empty
Downstream telemetry sees empty thinking text without a transport or parser error.
Agent Quick Fix
Repair against the current Claude API contract, then keep the change narrow and source-backed.
Product: Claude API
Current-contract area: Thinking summary silently disappears by default
Likely root cause: Downstream telemetry sees empty thinking text without a transport or parser error.
Repair direction: Verifier checks current model ID, accepted payload, structured output/tool parsing, and behavior under retries.Symptom
Downstream telemetry sees empty thinking text without a transport or parser error.
Why This Happens
SDK types may still accept fields that the selected hosted model rejects; model retirement, tokenizer, thinking, and tool behavior are external and fast-moving.
Common Wrong Fixes
- Changing local code without checking the current external contract.
- Retrying the same install, build, or API call with no version/source change.
- Applying a broad unsafe bypass when a narrow compatibility fix is available.
Codex Search Keywords
These are the search terms observed in a neutral Codex validation run for this failure shape.
site:docs.anthropic.com Claude API thinking summary default disappears thinking text empty
Claude API thinking summary extended thinking redacted_thinking thinking block text empty
site:docs.anthropic.com/en/docs/build-with-claude extended thinking thinking summaries Claude API
https://docs.anthropic.com/en/docs/build-with-claude/extended-thinking