Before setting task→review, coder agents should produce a mini evidence artifact: a short structured doc showing what was actually built, commands run, and key outputs/screenshots. Modeled on Showboat (https://github.com/simonw/showboat) but baked into the coder skill workflow. Addresses the trust gap where agents claim success at max iterations without actually completing work. The artifact should be committed alongside the code or attached as a task comment.
Pipeline: verification failed: Build failed for skills/coder.md (exit 1): 7[10000;10000H
[0m[38;5;1m[2Kfail: bild: nothing to build [0m[0m [0m
Pipeline scheduler: started run=pipeline-skills-coder-md-t-645-1771562441 domain=skills/coder.md
Pipeline scheduler: run=pipeline-skills-coder-md-t-645-1771562441 domain=skills/coder.md status=done cost=34c (fund-spend=failed)
Ava triage: pipeline auto-run reached status=done but the agent made NO git commits and reported blockers (missing files, path mismatches, or need clarification). This task is not actually in review — there's nothing to review. Resetting status to Open so it can be re-scoped.
Pipeline: dev completed (run=dev-t-645-1771537476, cost=0.0c)