AI API receipt · sgl-project/sglang
sgl-project/sglang#43153 · upstream bug reproduced by running the issue's own reproduction
Badgr cannot tell this response is wrong
A boolean argument written "true " (trailing space) came back as false.
false is a perfectly valid boolean; the schema is satisfied and Badgr cannot know the model said true. If the caller supplies the expected value (tool_arguments dry_run == true), Badgr rejects it: tool argument differs from the expected value. The receipt therefore records success. This is a limit of what the contract can prove, shown here so it is not mistaken for protection.
The output the bug produces, taken from the issue’s reproduction and fed to Badgr’s validators. Not a live provider call.
tool_calls: [{"type": "function", "function": {"name": "sql_interpreter", "arguments": "{\"dry_run\": false}"}}]
tools: ["sql_interpreter"]
tool_choice: "required"Written by Badgr’s receipt ledger and read back from GET /v1/receipts/replay-sglang-43153-boolean-flip.
{
"request_id": "replay-sglang-43153-boolean-flip",
"status": "success",
"endpoint": "/v1/chat/completions",
"stream": false,
"failure_stage": null,
"error_code": null,
"error_message": null,
"retryable": null,
"provider_mode": "replay",
"routing_decision": "replayed_upstream_output",
"created_at": "2026-10-09T02:14:36Z"
}Recorded 2026-10-09 in Badgr’s local development environment. Replay cases and verdicts are pinned by backend/test_ai_api_replay_cases.py.
← All AI API receipts