START 2026-10-27 14:25:38 IST
$ CURSOR_API_KEY_FILE=~/.secrets/cursor-api-key agent-sdk eval --dir . --no-stream --junit out/junit.xml --artifacts out/run-junit
[bdk] GitHub auth: unavailable (set GITHUB_APP_ID + GITHUB_APP_PRIVATE_KEY + GITHUB_APP_INSTALLATION_ID, or GITHUB_TOKEN / GH_TOKEN, or run `gh auth login`)
08:55:42.409 INFO  LocalCursorRulesService load completed meta={durationMs: 60, ruleCount: 0}
08:55:42.411 INFO  LocalCursorRulesService load completed meta={durationMs: 34, ruleCount: 0}
08:55:42.424 INFO  AgentSkillsCursorRulesService load completed meta={durationMs: 70, ruleCount: 12, skillCount: 1, servedFromCache: false}
08:55:42.438 INFO  AgentSkillsCursorRulesService load completed meta={durationMs: 58, ruleCount: 12, skillCount: 1, servedFromCache: false}
[session] tool lookup_benchmark start
[session] tool lookup_benchmark done (0ms)
[session] tool lookup_benchmark start
[session] tool lookup_benchmark done (0ms)
08:55:47.844 INFO  LocalCursorRulesService load completed meta={durationMs: 16, ruleCount: 0}
08:55:47.848 INFO  AgentSkillsCursorRulesService load completed meta={durationMs: 17, ruleCount: 12, skillCount: 1, servedFromCache: false}
[session] tool lookup_benchmark start
[session] tool lookup_benchmark done (0ms)
[session] tool lookup_benchmark start
[session] tool lookup_benchmark done (0ms)
PASS  benchmarks/swe-tool (7.4s)
  ✓ succeeded
  ✓ calledTool(lookup_benchmark)
  ✓ check(includes)
PASS  benchmarks/gpqa-json (6.8s)
  ✓ succeeded
  ✓ calledTool(lookup_benchmark)
  ✓ check(matches)
PASS  benchmarks/unknown-judge (7.8s)
  ✓ succeeded
  ✓ calledTool(lookup_benchmark)
  ✓ judge.closedQA(says it has no data on this benchmark and does not invent d…) [soft] (score=1.000 >= 0.8, Y (1.00) — The answer states there is no data on FooBarBench in the catalog and only describes lookup attempts, without inventing benchmark details.)

3 passed, 0 failed, 3 total
artifacts: ./out/run-junit
exit=0
wall_ms=15589
END 2026-10-27 14:25:53 IST