{"ok":true,"site":{"host":"agent-harness-lab.xiaoweiwan-michi.chatgpt.site","url":"https://agent-harness-lab.xiaoweiwan-michi.chatgpt.site/","type":"live","desc":"A shared reliability workbench for testing changes to an AI agent harness.","tools":[{"name":"get_lab_state","kind":"answer","impl":"imperative","description":"Read a bounded summary of the selected mission, stable revision, completed runs, staged candidate, decision, and recommended next agent actions.","title":"Get lab state","annotations":{"readOnlyHint":true,"untrustedContentHint":true},"inputSchema":{"type":"object","properties":{},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"load_mission","kind":"act","impl":"imperative","description":"Load one of the four built-in deterministic harness failure missions into a clean local workspace.","title":"Load mission","annotations":{"readOnlyHint":false,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{"mission_id":{"type":"string","enum":["completion","handoff","retry","authority"],"description":"Stable ID of a built-in mission."},"request_id":{"type":"string","minLength":1,"maxLength":64,"pattern":"^[A-Za-z0-9._-]+$","description":"Optional retry key. Reuse only for the same command and arguments."}},"required":["mission_id"],"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"run_baseline","kind":"act","impl":"imperative","description":"Replay the original harness against the selected deterministic target fixture and record its expected invariant failure in the visible workspace.","title":"Run baseline","annotations":{"readOnlyHint":false,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{"request_id":{"type":"string","minLength":1,"maxLength":64,"pattern":"^[A-Za-z0-9._-]+$","description":"Optional retry key. Reuse only for the same command and arguments."}},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"inspect_trace","kind":"answer","impl":"imperative","description":"Read up to three ordered, observable fixture facts from a completed baseline or candidate target run. Use offset to page without requesting the full trace.","title":"Inspect trace","annotations":{"readOnlyHint":true,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{"run":{"type":"string","enum":["baseline","candidate"],"description":"Completed target run to inspect."},"offset":{"type":"integer","minimum":0,"maximum":100,"description":"Zero-based fact offset. Defaults to 0."},"limit":{"type":"integer","minimum":1,"maximum":3,"description":"Fact count from 1 to 3. Defaults to 3."}},"required":["run"],"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"stage_harness_patch","kind":"act","impl":"imperative","description":"Stage the selected mission's fixed fixture patch with an optional causal hypothesis. This tool cannot change the declared diff, promote, or deploy.","title":"Stage harness patch","annotations":{"readOnlyHint":false,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{"hypothesis":{"type":"string","minLength":1,"maxLength":280,"description":"Causal explanation for why the fixed patch should change behavior."},"request_id":{"type":"string","minLength":1,"maxLength":64,"pattern":"^[A-Za-z0-9._-]+$","description":"Optional retry key. Reuse only for the same command and arguments."}},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"run_candidate_suite","kind":"act","impl":"imperative","description":"Run the staged candidate against the selected mission's target and two sealed local fixtures, then publish the derived five-signal comparison.","title":"Run candidate suite","annotations":{"readOnlyHint":false,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{"request_id":{"type":"string","minLength":1,"maxLength":64,"pattern":"^[A-Za-z0-9._-]+$","description":"Optional retry key. Reuse only for the same command and arguments."}},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"compare_harnesses","kind":"act","impl":"imperative","description":"Read the bounded five-signal comparison for the completed baseline and candidate suite, including sealed results, risks, and fixture limitations.","title":"Compare harnesses","annotations":{"readOnlyHint":true,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"},{"name":"export_evidence_receipt","kind":"act","impl":"imperative","description":"Build and validate the formal receipt, then return its bounded digest-bearing summary. This read does not download, promote, reject, or deploy anything.","title":"Read receipt summary","annotations":{"readOnlyHint":true,"untrustedContentHint":false},"inputSchema":{"type":"object","properties":{},"additionalProperties":false},"hasHandler":true,"hasExecute":true,"__source":"spec","executable":true,"handlerField":"execute"}],"category":"AI & Agents","favicon":"/favicons/agent-harness-lab.xiaoweiwan-michi.chatgpt.site.png","apiSurface":"spec","toolCount":8}}