McRitchie Studio

QA verdict: BLOCK (CI red). Core fix in bin/agent-worktree is CORRECT — verified the boundary math: the longest minted identifier (the _development_ name) lands at exactly 63 (never 64) for both mcritchie-studio and the shorter turf-monster prefix (max_len = 63 - prefix.length); fitting slugs pass through byte-for-byte (no churn); long slugs get a deterministic 8-hex SHA256 suffix; distinct long slugs sharing a 25-char prefix do NOT collide; dev/test share an identical base so db:test:prepare resolves. The 3 unit tests PASSED on CI. The block is ONE broken test, not the fix. DEFECT: the new [integration] 'db:test:prepare provisions a findable long-slug test DB' (test/commands/agent_worktree_test.rb ~L531-553) hardcodes credential-less connection strings — DATABASE_URL/TEST_DATABASE_URL = 'postgresql://localhost/<db>' and bare psql/dropdb with psql_env={PATH only}. On CI Postgres requires a password (CI sets DATABASE_URL=***localhost:5432), so the subprocess dies with 'fe_sendauth: no password supplied' BEFORE exercising the fix → 1 failure / 1565 runs (run 28146111302). It passes locally on trust-auth PG, which masked the gap. FIX: derive host/port/user/password from the CI-provided ENV['DATABASE_URL'] (URI.parse, swap only the database-name path segment) for both the db:test:prepare child env AND the trailing psql/dropdb (pass PGHOST/PGPORT/PGUSER/PGPASSWORD or a full conninfo) — or skip when ENV['DATABASE_URL'] is blank, but prefer deriving so the regression actually runs on CI. Re-run until the test job is green. Also update the stale claim — PR body + devops.checks_run say '[suite] ... 31 runs green' / 'CI green', but CI is red; correct that on resubmit. No rebase needed: branch is 0 behind origin/release with a clean merge-tree. PR left as DRAFT.

Post-merge review sign-off (cascade finished after the parallel conductor merged this): carl[heavy] APPROVE (ensure-adopt half-state guard proven red-before-fix via source mutation; preflight catches wrong-branch AND dirty-tree; overlap planner degrades gracefully), jasper[light] APPROVE. Non-blocking fast-follow: ship preflight is too strict on UNTRACKED files — git status --porcelain counts ?? entries, so the untracked retro-rel-*.md on the primary checkout would abort the NEXT ship even though --ff-only is safe with non-colliding untracked files. Exclude untracked (or downgrade untracked-only to a warning; keep tracked-modified + wrong-branch as hard aborts); fix-text should say 'stash -u'/'git clean'.

Post-merge review sign-off (cascade finished after the parallel conductor had already merged this onto rel-20260624-6f6638): carl[heavy] APPROVE (full regex probe — BARE_SEED_TASK has no hole; lookbehind/lookahead kill mydb:seed/db:seed_fund), shannon[light] APPROVE. Non-blocking follow-ups: (1) exempt-kind + zero-diff task hits exit 0 before the gate — move the post_deploy check ABOVE the exempt early-exit to close it; (2) db:setup/db:reset/db:prepare also seed but are out of this task's scope.

Review cascade (my session): steffon[heavy] APPROVE — built_by precedence correct (actor-soul>keep-existing>agent_slug, UUIDs rejected), busy exclusion sound (--busy + --busy-auto rescues SystemExit+StandardError), starve guard mathematically correct (never below MIN_CANDIDATES), seed reproducibility byte-for-byte (excluded_busy folds in only when non-empty). alex[light] APPROVE — docs accurate, no stale --builder-by-hand wording. 2/2 APPROVE, CI green, MERGEABLE. READY TO MERGE — no re-review needed. HELD per operator: parallel conductor (rel-20260624-6f6638) owns the merge/deploy lane.

Block resolved (9d262fb): color: mascot.signature_color added to all 4 MascotAgent.new sites (271 stage_timeline + 162 + the 2 already-correct), so the consolidated timeline mascot matches the build board. Regression test RED->GREEN (name-palette #3B82F6 -> signature #6F35FC); target suites 133 runs 0F. dor-check ✓. carl[deep] APPROVE stands. Merge-gated on CI test green.

shannon[heavy]: acceptance #7 partial miss — stage_agents_helper.rb:271 (stage_timeline) builds MascotAgent without color:, so the mascot on the /tasks/:id consolidated timeline shows the name-palette color while the build board shows the type (signature) color — same mascot, two colors. REQUIRED: add color: mascot.signature_color at line 271 (and line 162 for consistency; harmless there but align). Add a regression test asserting the consolidated-timeline mascot avatar carries the type color (mirror the build-board test). Non-blocking (defer): text-white low-contrast on light type colors (Electric/Ice/Steel/Fairy/Flying). Everything else approved.

Review cascade: carl[deep] APPROVE (gem 0.9.0 lock clean, migration matches engine reference, model sound, turf unaffected, seed correctly narrow via enumerals:seed); shannon[heavy/UI] BLOCK.

Review cascade: shannon[heavy] APPROVE (all 4 criteria met, clean revert of #163's crew-demotion bits while keeping operator-wanted gem/label/chip changes, no dead :strip/2xs/STRIP_CAP, theme+mobile parity improved via alpha mask-image) + carl[light] APPROVE (scope contained, dead-code cleanup complete, CI green). Non-blocking: stale :stack flex-wrap docstring (now grid). CI green.

Rework verified (7c270b03): board_mascot rewritten to read via subprocess (`task field <slug> mascot 2>/dev/null`, shellescaped) mirroring resolve_mascot — SystemExit from a non-2xx board read is now contained in the child, can never abort bin/task. + board-fallback regression test (fail_get knob → non-2xx → asserts exit 0 + marker written + no mascot). RED->GREEN (1 fail → 32/115 + 6/9). dor-check ✓. carl[heavy] block resolved; shannon[light] had approved. Merge-gated on CI test green.

Block resolved (cd3848b): bin/reviewer-select:13 stale ref alex-docs->alex (verified, comment-only). carl[light] APPROVE + alex-docs[heavy] block addressed. dor-check ✓. Merge-gated on CI test green.

Review cascade: shannon[heavy] APPROVE (migration safe/reversible/indexed, duration-spine protected, no N+1, ui+db coverage complete) + carl[light] APPROVE. Non-blocking nits deferred (kind filters in reviewer_selector/retro). CI green.

Review cascade: shannon[heavy] APPROVE (ranking deterministic, filter correct, no N+1, no traps, ui+db coverage complete) + carl[light] APPROVE. Non-blocking nits deferred (id-tiebreak, board_apps scoping). CI green.

Review cascade: shannon[heavy] APPROVE (all 4 criteria met in markup, no logic regression, no ERB/Alpine traps, theme+mobile parity) + alex-docs[light] APPROVE. Non-blocking nits deferred (dead :stack branch, STRIP_CAP comment). CI green.

Review: carl[light] APPROVE (migration reversible/idempotent, no broken refs); alex-docs[heavy] BLOCK on one stale ref — bin/reviewer-select:13 still says 'alex-docs=Documentation'. Fixing in-place (comment-only).

carl[heavy]: board_mascot calls api(:get) in-process; on a non-2xx board read api->die!->exit 1 raises SystemExit, which is NOT StandardError, so it escapes board_mascot's rescue AND write_feature_marker's rescue — aborting the whole bin/task move/create with exit 1 AFTER the server op succeeded (false failure; churn under the rate-limit epic). REQUIRED: make board_mascot incapable of terminating the process — mirror bin/agent-worktree resolve_mascot (shell out: `bin/task field <slug> mascot 2>/dev/null` in a subprocess) OR rescue SystemExit/Exception->nil around the api call. ADD an integration test: board-fallback path returns non-2xx -> command still exits 0 AND writes the marker. (Also: checks_run says unit(7); actual 6 tests/9 assertions.)

Review cascade: shannon[light] APPROVE; carl[heavy] BLOCK.

Excluded from rel (2026-06-24 QA release). CI fully red: Gemfile requires studio-engine (~> 0.9) but Gemfile.lock still pins 0.8.0, and frozen/deployment bundle refuses to update because studio-engine 0.9.0 (the Enumeral gem) is not published to RubyGems. Ordering: publish studio-engine 0.9.0 (eng#8) first, then bundle update studio-engine in mcr#157 to relock 0.9.0, push, confirm CI green + un-draft. Then resubmit.

Review cascade: carl[heavy] APPROVE — no residual fail-open (full gate set verified: contests/entries/wallets/cdp all route through geo_blocked?), blank-US branch reachable, no over-blocking (CO + non-US pass), kill-switch intact, current is find_or_initialize_by so no stale-read. shannon[light] APPROVE — kill-switch holds, CO positive control real, scope tight. Non-blocking nits (deferred): reorder geo_blocked? cheap-checks before enforcing? DB hit; add non-US+blank test.

alex-docs BLOCK resolved by rework (commit 195c71f): release.rb comment reworded to label the txn wrapper as defense-in-depth vs a distinct never-observed mode (verified on PR head); + elsif clarifying comment + no-op-without-reopen regression test. shannon[heavy] had APPROVEd the logic. Review satisfied; merge gated on CI test job green + PR marked ready.

Review cascade: shannon[heavy] APPROVE — reconcile breadth (!= assembled) is CORRECT, do not tighten (tightening makes adopt! inconsistent: new blocked task raises, existing blocked member silently no-ops); atomicity/idempotency/tests sound. alex-docs[light] BLOCK — release.rb 'Atomic:' comment misattributes the incident half-state to the txn wrapper (incident is fixed by adopt! reconciliation; the txn guards a distinct never-observed mode). Rework: comment fix + elsif clarifying comment + no-op-without-reopen regression test.