Agents Builders

Harden test loop and seed

Archived
harden-test-loop-and-seed

Created

Jun 24, 18:11

Started

Jun 24, 18:11

Completed

Jun 24, 18:59

DevOps handoff

Type

Feature

Shape

backend

Worktree Slug

harden-test-loop-and-seed

Repositories

mcritchie-studio

Release Train

Branch

feat/harden-test-loop-and-seed

Local URL

QA URL

Production URL

test-infra,dev-tooling,seeds

Acceptance Criteria

  • Tests run single-process locally, parallel only in CI
  • agent-worktree clears orphaned test procs before running
  • Board seed includes gem, crew, and qa-feedback cards

Expected Test Plan

  • [unit] worker-count logic; [integration] seed produces card states

Checks Run

  • [review] Carl (heavy/backend) APPROVE + Shannon (light) APPROVE — full suite single-process 1453 runs 0 fail; orphan-kill double-gated (pgrep+lsof cwd), seeds idempotent (re-run test green); CI parallelism preserved

Agent Context

Closes two retro frictions from the task-card sessions. #1 TEST INFRA: this env's parallel minitest workers fork-clone the test DB (pg fork-safety) and intermittently deadlock/segfault; orphaned rails-test procs from a killed run then hold the test DB and hang the next run. Fix at the source: test_helper.rb defaults workers to 1 LOCALLY, :number_of_processors only when CI is set, PARALLEL_WORKERS still overrides (matches the existing bin/agent-worktree single-process decision + comment). Single-process also means a killed run leaks at most ONE proc, not 14. Belt-and-suspenders: run_worktree_tests kills orphaned 'rails test' procs whose cwd is this worktree (cmdline+cwd scoped, never the dev server) before running. #4 SPARSE SEED: db/seeds/52_tasks.rb gets a gem-release card (library shape), a full-crew card (TaskEvents across stages → the under-slug crew grid), and a qa-feedback card (Activity) so the board's visual states render without hand-seeding. Tests: [unit] TestParallelism.worker_count; [integration] board seed loads the gem/crew/activity states.

Stage Timeline

Who handled each stage, the time it took (measured), and the model / tokens / cost reported (best-effort) — plus who's on it right now. means the agent didn't report that metric.

  1. Created Designed
    N Nidoran♂
    Nidoran♂
    Model
    Duration
    Tokens
    Cost
    Completed Jun 24, 18:11 · 3 days ago
    api
  2. Designed Building
    N Nidoran♂
    Nidoran♂
    Model
    claude-opus-4-8
    Duration
    under a minute
    Tokens
    Cost
    Started Jun 24, 18:11
    Completed Jun 24, 18:11 · 3 days ago
    cli
  3. Building Submitted
    N Nidoran♂
    Nidoran♂
    Model
    claude-opus-4-8
    Duration
    9 minutes
    Tokens
    Cost
    Started Jun 24, 18:11
    Completed Jun 24, 18:20 · 3 days ago
    cli
  4. Submitted Reviewed
    C Carl
    Carl primary
    S Shannon
    Shannon light
    Model
    claude-opus-4-8
    Duration
    9 minutes
    Tokens
    Cost
    Started Jun 24, 18:20
    Completed Jun 24, 18:29 · 3 days ago
    cli
  5. Reviewed Assembled
    S Steffon
    Steffon
    Model
    Duration
    2 minutes
    Tokens
    Cost
    Started Jun 24, 18:29
    Completed Jun 24, 18:31 · 3 days ago
  6. Assembled Shipped
    A Avi
    Avi
    Model
    Duration
    28 minutes
    Tokens
    Cost
    Started Jun 24, 18:31
    Completed Jun 24, 18:59 · 3 days ago
  7. Shipped Archived
    8
    81fad33e-98b5-4075-8ed6-d0efe6cb5804
    Model
    claude-opus-4-8
    Duration
    1 day
    Tokens
    Cost
    Started Jun 24, 18:59
    Completed Jun 25, 21:16 · 1 day ago
    cli

Conversation

QA review feedback, agent handoffs, and follow-up notes for this task.

No task feedback has been recorded yet.

Sealed-bid sizing

Edit →

Alex (PM)

Avi (PO)

Dev

Actual