openclaw - 💡(How to fix) Fix [Bug]: OpenAI strict tool schema downgrade diagnostic repeats on nearly every request

Bug type

Behavior bug (incorrect output/state without crash)

Beta release blocker

No

Summary

OpenAI Responses strict tool schema downgrade diagnostics repeat on nearly every request for an Azure OpenAI Responses route when the visible tool inventory contains strict-incompatible schemas.

Steps to reproduce

1. Run an OpenClaw gateway/agent request through an Azure OpenAI Responses model with debug logging enabled for model transport/tool payload diagnostics.
2. Expose the affected built-in tool inventory where 13 tool schemas are not strict-compatible.
3. Observe the same strict schema downgrade diagnostic repeat across successive requests within seconds.

Expected behavior

The provider payload should still downgrade to strict=false when the tool inventory is not strict-compatible, but the debug diagnostic for the same provider/model/transport/tool-violation signature should not be emitted on every request.

Actual behavior

The gateway repeatedly logs the same downgrade diagnostic:

OpenAI responses tool schema strict mode downgraded to strict=false for azure-openai-responses/gpt-5.3-codex because 13 tool schema(s) are not strict-compatible

The local evidence captured repeats within seconds at log lines 227, 238, 247, 259, and 268, and the broader log had 2,068 matching lines.

OpenClaw version

NOT_ENOUGH_INFO

Operating system

NOT_ENOUGH_INFO

Install method

pnpm dev / gateway dev log capture

Model

azure-openai-responses/gpt-5.3-codex

Provider / routing chain

openclaw -> azure-openai-responses

Additional provider/model setup details

The observed route is an Azure OpenAI Responses route. Private deployment names, session identifiers, local paths, and scheduled job names are intentionally redacted from this public report.

Logs, screenshots, and evidence

Trace/proof:
- gateway-dev.log:227
  "OpenAI responses tool schema strict mode downgraded to strict=false for azure-openai-responses/gpt-5.3-codex because 13 tool schema(s) are not strict-compatible"
- gateway-dev.log:238, :247, :259, :268 repeat the same message within seconds.

Observed count:
- 2,068 lines matching "strict mode downgraded" in the affected log.

Impact and severity

Affected: OpenClaw gateway/agent runs using native OpenAI/Azure OpenAI strict tool shaping with a strict-incompatible visible tool inventory.

Severity: Medium. The requests can still proceed with strict=false, but provider-side validation is reduced for that inventory and debug logs are flooded with repeated diagnostics.

Frequency: Repeated on nearly every observed request in the affected log.

Consequence: Heavy debug noise makes real provider diagnostics harder to inspect and obscures whether a new tool schema incompatibility appeared.

Additional information

The current code path recomputes diagnostics and calls log.debug whenever strictSetting === true, resolved strict mode is false, and debug logging is enabled. There is no de-duplication for the same provider/model/transport/tool-violation signature.