Why I Built Inkspect: How Do I Know the AI Is Really Done?

说明与用途

作者创建了 Inkspect，以验证 AI 代理是否真正完成任务，通过审查发现、证据和修复来确保在继续下一步之前完成工作。

问题

When a coding agent claims completion, the output may still contain hidden issues that surface later, causing regressions and wasted effort.

结果

After integrating Inkspect, the team could detect that interrupted runs lost model IDs, add them to checkpoints, and confirm the fix with passing tests for failure, cancellation, and recovery scenarios.

工作流

• 1. Assign a task to the coding agent.
• 2. Agent completes the task and generates a report.
• 3. Inkspect reviews the report, identifies findings, and records evidence.
• 4. If needed, Inkspect applies fixes and re‑checks the code.
• 5. Once all findings are resolved, the agent receives a handoff to start the next task.

证据

An agent would finish a task, but more problems would appear when I checked the result.

来源

The engine recorded model IDs returned by the AI service. These labels help explain which models were involved in a review.

来源

The fix was to include the session’s recorded model IDs in each stage checkpoint.

来源

The test checked three cases. The saved test results show the difference: ... After the fix, all cases passed.

来源