Propose Ship Quick

Run the full minerva lifecycle end-to-end with main-model decisions in place of human gates. This skill is the lightweight fast-path sibling of minerva:propose-ship-auto: same phases, same delegations, but where propose-ship-auto convenes a consensus panel, the main model decides directly. It is a hybrid orchestrator — it delegates to minerva:ship and minerva:cleanup directly, and inlines the propose / work / review / promote / replan phases so the main model can adjudicate them inline.

The four orchestrators form a ladder by adjudication cost: minerva:propose-ship (human gates) · minerva:propose-ship-quick (main model decides) · minerva:propose-ship-balanced (one advisory reviewer) · minerva:propose-ship-auto (consensus panels). Reach for quick when the change is genuinely small and the main model's own judgment is enough; reach for auto when the work is ambiguous or high-stakes enough to want independent agents arguing it out.

The mechanism: at each strategic or tactical decision point, the main model decides directly and fast, grounded in context. Operational decisions (commit messages, PR bodies, file paths) are executed without ceremony, exactly as in propose-ship-auto. The escalation path is preserved: a fail-closed escalation predicate sends genuinely-undecidable decisions to the user, and never elides the post-divergence or completion-verification self-checks.

Usage

• minerva:propose-ship-quick "fix the off-by-one in the date picker" — start a new quick run with the inline description as the seed.
• minerva:propose-ship-quick — start with current-session chat context as the seed (only sensible if the chat already discussed what to build).
• minerva:propose-ship-quick --cleanup-only <date-slug> --retry=N — internal re-entry from the cleanup-gate wake-up loop. Skips phases 1–6 and re-runs Phase 7.

Pre-flight: in-flight work collision

Identical to minerva:propose-ship's pre-flight. This check is not main-model-decided — a wrong call here destroys real work, so escalation to the user is hardcoded.

Read plugins/minerva/skills/propose/references/in-flight-check.md and run it. It reads four evidence sources — local work units (via the in_flight predicate, never a string match), local and remote branches, open PRs, and live sibling Claude sessions — each failing soft, so a repo with no remote, no tracker and no siblings passes through silently. It is detection, not a lock: git worktree add -b serializes only sessions choosing the same slug, so a clean result means no evidence was found, not that nobody else is working the goal.

When a peer session messages you, read plugins/minerva/skills/propose/references/cross-session.md: inform, never delegate.

A collision is a hardcoded AskUserQuestion (resume that work / start fresh anyway / abandon this run) and increments the global escalation counter.

Only proceed after the user confirms. This is the single mandatory — and only guaranteed — pre-run user interaction; everything else reaches the user only via the escalation predicate.

Solo-decision protocol

The full policy — the default (main model decides), the fail-closed escalation predicate, the scope-fit escape, the never-bypassed self-checks, the hardcoded escalation triggers, the escalation counter, and per-decision logging — lives in references/solo-decision-protocol.md. Read that file once, in full, before this run's first strategic/tactical decision point; its rules then apply to every decision that follows.

Binding floor, even before the reference is read:

• The main model decides each strategic/tactical decision directly. It does not convene a minerva:round-table panel — that is the whole point of this skill, and the one behavior that distinguishes it from propose-ship-auto.
• Before deciding, the main model applies the escalation predicate silently per-decision. It fails closed to the user: on genuine ambiguity (no dominant option after honest analysis), high blast-radius / irreversibility, an unfamiliar public interface or cross-cutting contract, a conflicting .minerva/knowledge/ constraint — it escalates instead of guessing.
• Never elided: completion verification, mid-work divergence confirmation, new-plan acceptance (replan), and every hardcoded escalation trigger. These are run as rigorous main-model self-checks regardless of how small the change looks; the model escalates on genuine uncertainty rather than rubber-stamping its own work.
• Scope-fit escape: if the change proves not small (scope explosion, deep complexity), escalate recommending a switch to propose-ship-balanced (one reviewer), propose-ship-auto (panels), or propose-ship (human gates).
• Every decision and escalation logs one line to the work unit's scratchpad.md under a ## Quick decisions YYYY-MM-DD header ([decided] / [escalated to user]).

Phases

Execute the phases in order. The full inline protocols live in references/phases.md. Before executing each phase, read that phase's section there; the map below locates the work, it is not the protocol:

1. Propose (inline) — assemble context → design synthesis → the main model decides scope, approach, and whole-proposal soundness (escalating on genuine uncertainty) → worktree + branch + file writes per minerva:propose → self-review.
2. Work (inline) — implement per minerva:work's protocol; on a suspected load-bearing divergence the main model confirms it itself (escalate if unsure); completion-verification self-check on the success-criteria checklist + diff.
   • 2.5 Replan (inline, if triggered) — draft Original plan / What changed / New plan; the main model accepts the new plan (escalate if unsure); append to replan.md.
3. Review (inline) — minerva audit + code review (PR mode delegates to code-review:code-review); the main model triages all findings; replan-vs-FIX decided by the main model if a load-bearing finding surfaces.
4. Promote (inline) — the main model partitions PROMOTE/MERGE/DISCARD/TODO and disposes TODOs; apply writes per minerva:promote Mode A; archive scratchpad.
5. Ship gate — no gate: silent advancement, except halt if the global escalation counter has reached 3.
6. Ship (delegated) — invoke minerva:ship via the Skill tool with its auto-mode instruction (auto-accept hard gates #1 commit message and #2 PR title/body; everything else unchanged). CI auto-fix bails classified other are escalated to the user — never silently decided.
7. Cleanup gate — poll PR state via gh pr view; on MERGED invoke minerva:cleanup via the Skill tool with args <date-slug> --yes; on OPEN with auto-merge, ScheduleWakeup re-entry (--cleanup-only <date-slug> --retry=N, cap 12); otherwise surface manual instructions.

Failure modes, escalation, budget caps

Binding caps: propose-phase abort when the propose-phase decisions escalate to the point the strategic intent is too ambiguous for the main model to resolve; global escalation counter halts the run at 3. Hard escalation triggers that skip the main model's judgment entirely: in-flight collision, worktree-creation failure, ship-phase failures (other classification, push rejection, gh auth failure), counter at 3. The full trigger list, final-report-on-bail format, and observability requirements live in references/governance.md — read it at the first escalation or before reporting any bail.

Out of scope

Never modify any existing minerva skill at run time (this skill orchestrates by invocation only); never auto-cascade into new work units; never cap implementation time; review/promote ordering is fixed. This skill never convenes a minerva:round-table panel. Rationale and detail: references/governance.md.