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-journey

Replay a StatePort Card's recorded Journey on retained source or current local code, inspect its steps and progress, and explain the terminal outcome.

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-journey

技能说明


name: stateport-journey description: Replay a StatePort Card's recorded Journey on retained source or current local code, inspect its steps and progress, and explain the terminal outcome. license: MIT

StatePort Journey

Use this skill to inspect or replay a recorded interaction sequence. Read the Journey procedure, workflow guidance and data handling.

Inspect the supplied Card and its Journey with inspect_state and inspect_journey. Use inspect_journey_action if the next recorded step helps explain the sequence or a failure. For inspection-only requests, report this evidence without starting a run.

For replay, select the target from the task: frontendOrigin for current loopback code, omitted for retained captured source. A localhost capture needs the running local frontend. Ask only if the target is genuinely ambiguous.

start_journey requires explicit user confirmation of this exact Card. An existing request to replay the named Card provides that confirmation; do not ask again. Set confirmedByUser: true only on that basis. Use allowProtectedLocalValuesForThisRun: true only when the user also authorized protected local values for this run. If that decision is missing and required, ask for it without exposing any values.

Read get_journey_progress until completion, a failure or a manual barrier. Avoid rapid polling. Use stop_run to finalize and retain the terminal result; honor an explicit request to leave the visible run open and explain that its final result is still pending. Do not stop early and report completion.

Report the target, completed/total actions, terminal outcome and useful next step for any failure. Read the persisted run/outcome and compare runs when the task includes verification. A Journey completed on retained source is evidence about that source; current-code verification needs a run on that code.