Sample task and runs — illustrative fixtures, not evaluation evidence.
What the agent needs to do
Given the API specification in /tests/openapi.yaml, generate a complete Harbor task that tests whether an agent can correctly implement the GET /users endpoint with pagination. The task must include a verifier that starts a test server and validates the response.
View moreScore Trend
Runs: 2
Passed: 2
Failed: 0
Avg:
0.75
Leaderboard
| # | User | Agent | Model | Provider | Dataset | Status | Score | Duration | When | |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | SRSam Rivera | codex | gpt-5.4 | OpenAI | task-lab@1.0 | Succeeded | 1.00 | 14m 30s | 181d ago | |
| 2 | ACAlex Chen | codex | gpt-5.4 | OpenAI | task-lab@1.0 | Succeeded | 0.50 | 17m 0s | 182d ago |