inspect-workflow

name	inspect-workflow
description	Use to get a quick health view of a workflow — its definition, registered versions, recent instances and their statuses, and which connections it touches. The "is this thing healthy / what does it look like right now" view, distinct from trace-instance which drills into one specific run.

Inspect a workflow

Pulls workflow definition and recent execution state from the runtime API (port 7001).

1. Find the workflow

By name (substring search):

curl -s "http://127.0.0.1:7001/api/runtime/workflows?search=<name>&pageSize=20" \
  | jq '.items[] | {id, name, path, currentVersion, updatedAt}'

By folder path:

curl -s "http://127.0.0.1:7001/api/runtime/workflows?path=<folder>&recursive=true&pageSize=100" \
  | jq '.items[] | {id, name, path}'

List everything:

curl -s "http://127.0.0.1:7001/api/runtime/workflows?pageSize=100" | jq '.items | length'

2. Get the definition

Latest version:

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>" \
  | jq '{id, name, currentVersion, definition, inputSchema, outputSchema}'

A specific version:

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>?versionNumber=3" | jq .

For just the step graph:

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>" \
  | jq '.definition.steps[] | {id, type, name}'

3. Recent instances

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>/instances?page=0&size=20" \
  | jq '.items[] | {id, status, startedAt, durationMs, error: .error.message}'

For status breakdown:

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>/instances?size=100" \
  | jq -r '.items[].status' | sort | uniq -c

For only failures:

curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>/instances?size=50" \
  | jq '.items[] | select(.status=="failed") | {id, startedAt, error: .error.message}'

Then drill into a specific failure with the trace-instance skill.

4. What connections does it use?

# extract integration_ids referenced by Agent steps
curl -s "http://127.0.0.1:7001/api/runtime/workflows/<WORKFLOW_ID>" \
  | jq -r '.definition.steps[] | select(.type=="Agent") | .config.connectionId // empty' \
  | sort -u

For each connection ID, see the inspect-connection skill.

5. Sanity checks

"Workflow not appearing in the list" → version probably hasn't been registered. Re-run the compile + register flow from e2e-verify step 5–6.

"Steps in the picker don't match what I see in the definition" → frontend client may be stale; re-run regen-frontend-api.

"All instances suddenly failing the same way" → check whether anything changed in: a connection (token expired — inspect-connection), an agent (rebuild + re-register), or a downstream service (out-of-band).