工具 API
list_ runs
List runs owned by the current credential (newest first, paginated); metadata only, no result data. Use it to: - recover a lost run_id (an async run keeps executing and billing on the platform after the session breaks; narrow by app / created_from, then continue with get_run_status); - check many runs at once (filter by status to see which are not terminal yet); - audit cost (filter by app / time range; see items[].billing). Ownership is per user: runs launched from the portal / SDK / API / MCP all appear here, **not only those launched in this session**. Returns items[] (run_id / state / created_at / usage / billing ...) and pagination (has_more=true: increase offset to page).
调用信息
- 工具标识
- datahub.list_runs
- 服务提供方
- 八爪鱼 DataHub
- 平均响应
- 571 ms
- 近 7 天调用
- 5
输入参数
statusFilter by run state; omit = all
appFilter by App reference `<publisher_username>/<app_name>`; an invalid reference yields an empty result set, not an error
triggered_byFilter by launch channel; omit = all
run_kindFilter by run kind: production / test (App author debugging); omit = all
created_fromLower bound (inclusive) of launch time, ISO-8601 instant, e.g. 2026-09-01T00:00:00Z
created_toUpper bound (exclusive) of launch time, ISO-8601 instant
offsetPagination start offset
limitItems per call, 1-50 (larger values are clamped)