fix(crews): unit stop and archive finish over a seat that was never created

[bryantderosier]

Important

⚠️ Merge order: merge these in this exact order

This PR is part of the Crews stack. Merge top to bottom, one at a time, and let each land on j5/main before the next. #292 isn't in the stack but has to merge before #313.

1. fix(crews): Crew notices report only what the platform measured #292: Crew notices report only what the platform measured (adds migration 020). Not stacked, but it must merge before fix(crews): a proposal launches once and reports every seat #313.
2. fix(crews): no seat launches into a retired Crew or under a gone Captain #271: No seat launches into a retired Crew or under a gone Captain
3. fix(crews): unit stop and archive finish over a seat that was never created #279: Unit stop and archive finish over a seat that was never created ⬅️ this PR
4. fix(crews): a proposal launches once and reports every seat #313: A proposal launches once and reports every seat (adds migration 021)
5. fix(crews): Crew groups keep seats without thread facts as unknown #300: Crew groups keep seats without thread facts as unknown
6. fix(crews): a Crew follows its Captain through every lifecycle step #315: A Crew follows its Captain through every lifecycle step (adds migration 022)

Why #292 goes first: J5 migrations run in id order, and the migrator skips any id at or below the newest one a database has already applied. If #313's migration 021 ships before #292's 020, 020 never runs on that database.

A launch records every seat's row before any seat thread exists, so a launch or addition that failed partway leaves rows with no thread behind them. Unit stop and unit archive read each row as a live seat and failed on the first missing one: a stop gave up before it reached later seats that were still running, and the Crew couldn't be archived as a unit.

What I changed:

• The proof of "never created". A seat qualifies only when two absences agree: no home is registered for its deterministic thread id, and the projection store answers not-found for it. Both reads run under the per-Crew lock from fix(crews): no seat launches into a retired Crew or under a gone Captain #271, so no launch or addition is creating the seat meanwhile. A store that can't answer, or a home with no thread behind it, still fails the operation loudly.
• Unit stop takes the same lock, reports a missing seat as never_created, and still interrupts every running seat after it.
• Unit archive (archive_crew and the Fleet's Archive crew) skips such seats, reports them as never_created, finishes on the seats that exist, and stamps the Crew retired. The row stays on the retired roster. Under the product(j5): decide what archive, unarchive, and settle mean for agents, Crews, and their children #254 ruling these doors stay as batch conveniences, and even with launch-once dropping failed-create rows, a restart mid-launch still leaves rows with no thread.
• Wire shape. The HTTP routes leave never-created seats out of members, so every client decodes what it gets. The agent-facing MCP results say never_created per seat.

What this doesn't cover: a crash after a seat's thread commits but before its home registers still stops unit archive with a mismatch error. That's the loud failure I want there, since the seat exists, but it means not every partial Crew archives cleanly.

Tests:

• ArchiveAgentService.test.ts: readFacts returns null only when both the home and the thread are absent; an unreadable store and a home with no thread both fail.
• ArchiveCrewService.test.ts: the unit retires past a never-created seat and replays as already archived; a seat whose facts can't be read fails the unit and retires nothing.
• CrewStopService.test.ts: a missing seat ahead of a running one still lets the stop interrupt it; an unreadable seat or a homed seat with no thread fails the stop.
• CrewLaunchService.test.ts: with the real launcher, an addition reserves its seat, the spawn fails, and the unit archive still finishes.
• CrewArchiveHttp.test.ts, CrewStopHttp.test.ts: never-created seats are left out of members.

One upstream-file touch: the existing J5 Crew-stop block in orchestration-v2/runtimeLayer.test.ts gains two mock lines (the lock and the home read). #232 already tracks recording that block in FORK.md.

Stacked on #271.

Closes #224

Done by Claude Opus 5.5 in Claude Code (T3 Code).

🤖 Generated with Claude Code

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: Repository: Jacksondr5/j5code/.coderabbit.yaml

Review profile: CHILL

Plan: Advanced

Run ID: c82f5ed4-5635-4d30-a184-431b541065bd

---

_{Comment @coderabbitai help to get the list of available commands.}

[Jacksondr5]

Claimed for review by Jackson with Claude Opus 5.5 and GPT-6-Astra (the crews review thread). Other agents: please skip this one.

[Jacksondr5]

Covered together with #271 in one comment, since the question is the same for both: #271 (comment)

The short version for this PR: the unit-stop handling survives option A and is worth landing on its own. The unit-archive and Captain-cascade skipping hardens code your cascade-removal PR takes out, and the main case (a row left by a failed create) goes away when launch-once drops that row. Two smaller notes if it lands as is. Nothing reads the new neverCreatedSeats field yet. And the upstream orchestration-v2/runtimeLayer.test.ts touch waits on #232; moving that Crew-stop test into a J5-owned file would avoid the upstream edit entirely.

Reviewed by Claude Opus 5.5 in Claude Code, with an independent pass by GPT-6-Astra in Codex.

[bryantderosier]

Replied on #271 for both PRs, since the question was the same: #271. The short version for this one: I dropped neverCreatedSeats and the cascade test in 6976480, and fixed the body to name the thread-committed-but-home-missing case it doesn't cover. Unit archive's never-created handling stays, because archive_crew and the Fleet's Archive crew survive option A and a restart mid-launch still leaves rows with no thread. The runtimeLayer.test.ts touch stays, and I explain why on #271.

[Jacksondr5]

Approving at 6976480. neverCreatedSeats and the cascade test are gone, and the body now names the thread-committed-but-home-missing case it doesn't cover. You're right that unit archive keeps its never-created handling: archive_crew and the Fleet's Archive crew survive option A, and a restart mid-launch still leaves rows with no thread. The runtimeLayer.test.ts touch is the smaller cost than duplicating the upstream orchestrator wiring; recording it under #232 is the right place. Merges after #271.

Reviewed by Claude Opus 5.5 in Claude Code, with an independent pass by GPT-6-Astra in Codex.