跳至正文
OpenAgentSkill
浏览已开放 · 目录记录日期 2026-11-21 · 自动更新和账户服务暂时关闭。
开发与测试

stateport-replay-experiments

Investigate StatePort replay mismatches or test frontend behavior with request remaps, replay/block/local-live routes and JSON response overlays, then compare or restore the baseline.

StatePort-Dev · 已有静态检查记录

安装与使用

运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。

npx skills add StatePort-Dev/stateport-agent-skills --skill stateport-replay-experiments

技能说明


name: stateport-replay-experiments description: Investigate StatePort replay mismatches or test frontend behavior with request remaps, replay/block/local-live routes and JSON response overlays, then compare or restore the baseline. license: MIT

StatePort replay experiments

Help test the user's hypothesis using the existing Card and supported replay controls. Read the experiment procedure, workflow guidance and data handling.

Inspect the Card and intended local frontend. Use an existing task-owned session or open the Card on that frontend. Read list_routes, inspect_exchange and, for changed requests, inspect_request_changes. Choose the relevant route or trusted evidence from these results; ask only when multiple choices would test different hypotheses and the task does not settle it.

Use the operation that answers the question:

  • apply_request_remap / remove_request_remap for eligible captured-response remaps using runtime-issued evidence IDs.
  • set_route_mode for replay, block, or explicitly requested Live routing to the selected loopback backend.
  • create_response_overlay / remove_response_overlay for JSON merge-patch experiments and returning to the baseline.

Carry out the authorized experiment without asking before each ordinary step. If testing would require an unspecified backend, a persistent revision overlay or an external action outside the task, clarify that choice first. Default a remap to current_experiment unless persistence was requested.

Retain baseline and candidate run IDs, follow replacement session IDs after changes, then finalize and compare runs. Record the routes, patch or remap that changed the conditions. Restore the baseline when the task is complete unless the user asked to retain the experiment. Some Card types cannot use overlays; report the runtime result and offer an applicable alternative.

Explain the result in terms of the tested hypothesis. A patched or remapped captured response can demonstrate frontend behavior; backend correctness requires evidence from the current backend.