weryai-task-inspector

# WeryAI Task Inspector Inspect existing WeryAI tasks and batches through the official task query endpoints. This skill is read-only: it never submits new jobs and never mutates remote state. ## Example Prompts - `Inspect this WeryAI task ID and tell me whether it is finished.` - `Query this WeryAI batch ID and summarize the state of each task.` - `Check whether this podcast task reached audio-success yet.` - `Show me the output artifacts for this WeryAI task, including audio or video URLs if present.` ## Quick Summary - Main jobs: `task status lookup`, `batch status lookup`, `artifact inspection`, `podcast content_status inspection` - Inputs: `taskId` or `batchId` - Main outputs: normalized status, best-effort artifacts, raw task payload - Main trust signals: official read-only task query endpoints, explicit no-submit boundary, podcast-aware `content_status` handling ## Prerequisites - `WERYAI_API_KEY` must be set before running task queries. - Node.js `>=18` is required. - This skill is read-only and safe to re-run. ## Preferred command ```sh node {baseDir}/scripts/task-inspector.js --task-id <task-id> node {baseDir}/scripts/task-inspector.js --batch-id <batch-id> ``` ## Workflow 1. Determine whether the user has a single `taskId` or a `batchId`. 2. Query the corresponding official endpoint. 3. Return normalized state plus best-effort extracted artifacts. 4. For podcast tasks, interpret `content_status` as the primary lifecycle signal. ## Output contract Successful output includes: - `ok` - `phase` - `queryType` - `taskId` or `batchId` - `taskStatus` - `contentStatus` when present - `artifacts` - `raw` Failure output includes: - `errorCode` - `errorMessage` ## Definition of Done - A valid `taskId` returns normalized task state plus best-effort artifacts and the raw payload. - A valid `batchId` returns a batch-level phase and per-task normalized state. - Podcast tasks surface `contentStatus` when present. - Invalid or inaccessible task identifiers fail with a clear `errorCode` and `errorMessage`. ## Re-run Behavior - Re-running this skill is read-only and safe. - It never creates new WeryAI tasks and never spends credits by itself. ## Boundaries - Do not use this skill to create new image, video, music, chat, or podcast jobs. - Do not invent undocumented output fields when the raw payload does not contain them. - Do not treat podcast `task_status=completed` as final if `content_status` has not reached `audio-success`. ## References - Official task query summary: [references/tasks-api.md](references/tasks-api.md) - Task details API: [Query Task Details](https://docs.weryai.com/api-reference/tasks/query-task-details) - Batch task API: [Query Batch Task Status](https://docs.weryai.com/api-reference/tasks/query-batch-task-status)