wisp-diagnose

WISP — Diagnose Run

Inspect a run's event timeline to figure out what failed and why.

Platform note: snippets below show both bash and PowerShell forms. Pick the one matching the user's shell. The bash form also runs from Git Bash / WSL on Windows.

Inputs

• runId: the run UUID. Ask the user if not provided.

Steps

1. Fetch the events:

   # bash
   curl -s "http://127.0.0.1:${WISP_PORT:-4400}/api/runs/<runId>/events?limit=500"
   

   # PowerShell
   $port = if ($env:WISP_PORT) { $env:WISP_PORT } else { 4400 }
   Invoke-RestMethod -Uri "http://127.0.0.1:$port/api/runs/<runId>/events?limit=500"
   

   Response: {"events": [{"type":"...", "payload":..., ...}, ...]} ordered oldest-first.

2. Show the timeline, condensed: print one line per event of interest. Skip noisy types (task.text-delta, task.usage — print usage totals at the end instead). Format:

   <timestamp>  <event type>  <task id>  <one-line summary>
   

3. Highlight failures: for each task.failed, harness.verify-failed, qa.replan-triggered, qa.replan-exhausted, rate-limit.hit, run.paused event, print the FULL payload (not a one-liner). Especially:

   • harness.verify-failed.payload.failures[*] — kind, cmd, exitCode, tail
   • harness.verify-failed.payload.output — full output (truncate to last 100 lines if huge)
   • qa.replan-triggered.payload.reason — what QA reported

4. Token & duration totals: query /api/runs/<runId> to get the per-task token/turn totals. Print as a small table.

5. Verdict: in 1-2 sentences, summarise WHY the run failed (or what it's currently stuck on). Quote a specific event payload as evidence.

Useful event types to know