colima-assignment procedure — revision 10f80d6e

Procedure colima-assignment · revision 10f80d6e · 2026-09-13

What this test proves

Inside the colima testbed, a real agent executes prompt AT10 and observes assignment transitions only through public ATM CLI JSON. Queue, ready, close, reassign, move, cancel, and busy-assigner behavior must match the task protocol while pending-ack remains untouched.

Evidence layout

This procedure is one step of a colima integration run. The agent's unedited prompt-report-1 JSON is retained beside step.json; scripts/integration/render_colima.py renders the panel and aggregate. Historical payloads remain byte-identical and render through the same path.

Changes

The revision sections below list every runner revision that produced committed evidence, newest first. A run whose source revision is not an ancestor of any listed revision links to the revision in effect on its run date and is marked inferred.

Flow

Start one testbed fixture

Run AT10 assignment agent prompt

1: queue and ready

2: close releases next

3: reassign

4: move head

5: cancel

6: busy assigner terminal lines

Aggregate PASS/FAIL

Write prompt report and step.json

runner revision 10f80d6e

Steps

stepactionobservableevidence
1Tell the agent to assign three taskstask events show queued positions 1–3, one ready, and pending-ack zero`prompt-AT10.json` and `step.json` cases
2Start and close the ready taskbounded task-list/events polling shows the next task ready`prompt-AT10.json` and `step.json` cases
3Reassign a queued taskpublic mail/events show reassigned to the old assignee and queued to the new`prompt-AT10.json` and `step.json` cases
4Move an idle assignee's queued task to headtask events show exactly one new ready transition`prompt-AT10.json` and `step.json` cases
5Cancel a queued taskpublic mail/events show the cancelled terminal outcome`prompt-AT10.json` and `step.json` cases
6Start and complete while the assigner is busyassigner mail/history shows started before completed`prompt-AT10.json` and `step.json` cases

Revision note: refactor(bb8): drive colima checks through fixture prompts