McRitchie Studio

← All Tasks

Sizing

Auto-commit release retro artifacts

Archived

autocommit-release-artifacts

Created

Jun 26, 04:24

Started

Jun 26, 04:39

Completed

Jun 26, 05:33

DevOps handoff

Type

Feature

Shape

backend

Worktree Slug

autocommit-release-artifacts

Repositories

mcritchie-studio

Release Train

—

Branch

feat/autocommit-release-artifacts

Pull Request

https://github.com/amcritchie/mcritchie-studio/pull/222

Local URL

—

QA URL

—

Production URL

—

backend deploy-tooling git

Acceptance Criteria

bin/release retro commits the retro doc to release
archive commits the delete-later ledger update
Guarded to main==release; best-effort, non-fatal
Stops preflight stash accumulation of generated docs

Expected Test Plan

[unit] Release::ArtifactCommit decision (porcelain parsing + guard)
[integration] release suite + release_flow green, no regression

Checks Run

[unit] artifact_commit_test: safe_to_commit?/other_dirty_paths (doc-only, ledger, rename, dirty)
[integration] release suite 215/0 green with retro+archive commit wired (no regression)
[full-suite@81f12e40441ab1178a02e5fb832ed108fc1124d7] bin/rails test green
[rubocop@81f12e40441ab1178a02e5fb832ed108fc1124d7] bin/rubocop clean

Agent Context

Root cause of the stash pile (see commit-backlogged-retros / PR #220): bin/release retro File.writes docs/agents/audits/retro-<slug>.md untracked (bin/release:1669) and archive leaves delete-later.md modified, so the ship preflight's clean-main requirement forces the conductor to stash them every ship. FIX: after retro writes the doc (and after archive updates the ledger), commit it to the release branch + push, MIRRORING the gem re-pin's proven checkout/commit/push pattern at bin/release:1188-1216. DESIGN/SAFETY: make it best-effort + non-fatal (rescue → leave uncommitted, today's behavior) and GUARD on local main == origin/release (the post-ship state these run in) so it never commits when main!=release or pushes a non-ff; on push failure, fall back to leaving the doc uncommitted (preflight stashes it, no regression). Since post-ship main==release, pushing the commit to release means the NEXT ship ff's main up to it — no divergence. TESTS: cover the guard (main==release commits+pushes; main!=release skips), the non-fatal rescue path, and that a generated retro doc ends up committed not stashed. Verify against the existing bin/release test harness (test/lib + test/commands). Edge: retro can run standalone for an old slug — committing that doc to release is still correct/additive.

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.

Sizing Avi · PO — Dev — Actual XL

Created → Designed

P
Pidgeot

Model

—

Duration

—

Tokens

—

Cost

—

Completed Jun 26, 04:24 · 1 day ago

api
Designed → Building

P
Pidgeot

Model

claude-opus-4-8

Duration

15 minutes

Tokens

—

Cost

—

Started Jun 26, 04:24

Completed Jun 26, 04:39 · 1 day ago

cli
Building → Submitted

P
Pidgeot

Model

claude-opus-4-8

Duration

9 minutes

Tokens

36,835,147

Cost

~$20.59

Started Jun 26, 04:39

Completed Jun 26, 04:49 · 1 day ago

cli
Submitted → Reviewed

C
Carl primary

S
Shannon light

Model

claude-opus-4-8

Duration

25 minutes

Tokens

14,790,150

Cost

~$11.46

Started Jun 26, 04:49

Completed Jun 26, 05:14 · 1 day ago

cli
Reviewed → Assembled

S
Steffon

Model

claude-opus-4-8

Duration

2 minutes

Tokens

2,388,927

Cost

~$1.67

Started Jun 26, 05:14

Completed Jun 26, 05:16 · 1 day ago
Assembled → Shipped

A
Avi

Model

claude-opus-4-8

Duration

17 minutes

Tokens

17,049,713

Cost

~$11.21

Started Jun 26, 05:16

Completed Jun 26, 05:33 · 1 day ago
Shipped → Archived

—

Model

—

Duration

about 13 hours

Tokens

—

Cost

—

Started Jun 26, 05:33

Completed Jun 26, 18:43 · about 15 hours ago