dart-ultrawork

Use this skill in Codex to run the DART dart-ultrawork workflow. The editable workflow source lives in .claude/commands/; this file is its generated adapter in the shared .agents/skills/ catalog.

Invocation

• Claude Code: /dart-ultrawork <arguments>
• Codex: $dart-ultrawork <arguments>

Treat the text after the skill name as $ARGUMENTS. When the workflow references $1, $2, etc., map those to the positional values supplied by the user.

Command Body

Start a team-scale or autonomous DART task: $ARGUMENTS

Required Reading

@AGENTS.md @docs/ai/principles.md @docs/ai/north-star.md @docs/ai/orchestration.md @docs/dev_tasks/README.md @docs/ai/verification.md

Load additional owners only when the matching phase needs them:

• placement or cleanup: docs/README.md;
• numbered-plan selection or packet state: docs/plans/dashboard.md;
• version control, changelog, tools, or review: docs/onboarding/{contributing,changelog,ai-tools}.md.

Arguments

$ARGUMENTS is a task brief plus optional mode flags:

• mode=interview: ask one up-front batch of critical questions.
• mode=brief: treat provided context as sufficient unless escalation applies.
• mode=resume: start from the existing docs/dev_tasks/<task>/ project home and run the session-start protocol before changing files.
• interview=skip: skip maintainer questions only when the brief already answers all consequential decisions.

The brief may be prose or a structured TASK / CONTEXT block. Extract north star, deliverable, acceptance criteria, constraints, risks, references, paths, issues/PRs/branches, commands, and first step when present.

Workflow

Own understanding, decomposition, sequencing, review, and honest evidence for the whole task. Follow the orchestrator/executor and packet-sizing contracts in docs/ai/orchestration.md. Delegate only when the user explicitly requested it and the current surface permits it; otherwise execute packets serially. Use dart-new-task for bounded single-session work unless the user asked for the autonomous project-home loop.

1. Session start and current reality - Follow docs/dev_tasks/README.md's Session Start protocol for the docs/dev_tasks/<task>/ project home: current snapshot and next action first, history only as needed, then verify live branch/PR/plan state before acting. Run pixi run ai-doctor when setup, discovery, instruction, agent, or hook state is uncertain. Create or refresh the project home before implementation when the session policy requires it.
2. Understand and scout - Restate the north star, final deliverable, acceptance criteria, quality bar, non-goals, constraints, risks, and target branch line (DART 7 main, DART 6 LTS, or both). Scout the territory first with named docs/code, read-only searches, a dart-analyze pass, the Codex dart_scout profile, or focused reference review; draft a candidate decomposition privately before asking anything.
3. Interview decisions; self-resolve uncertainties - Ask at most one up-front batch of critical questions, only for choices or authority missing from the brief and prior decisions. Escalate before destructive operations, history rewrites, irreversible migrations, meaningful cost, security/credential/secret handling, legal or privacy-sensitive decisions, major product-direction choices not covered by the brief, conflicts with stated constraints, or any assumption whose wrong answer could cause significant harm. If input is unavailable, choose the safest reversible path, document the assumption, and continue only with non-blocked work. Then split consequential unknowns:
   • Maintainer decisions: preference, scope, public API, release, quality-bar, or roadmap calls that evidence cannot settle. Ask the human now in one batched interview (focused questions with 2-4 concrete options each, recommendation first). Defer work that depends on an open decision; continue independent work already authorized. Skip this discretionary interview when mode=brief; also skip when interview=skip and the prompt already answers everything consequential. In both cases, still follow the escalation rules above.
   • Evidence-resolvable uncertainties: anything a focused A/B test, benchmark, throwaway spike, reference lookup, or blind-spot review can settle. Do not ask the human; schedule these as spike/research packets and record the method and result as evidence (see "Discovering unknowns before committing" in docs/ai/orchestration.md).
4. Create or refresh the tracking surface - Populate the project home with value, north star, deliverable, scope, non-goals, assumptions, risks, acceptance evidence, gates, dependencies, milestone, next actions, and blockers. Claim-dependent 3D structure or behavior work routes through dart-verify-sim. Keep RESUME.md as the handoff; add decisions.md, verification.md, and progress-log.md sidecars when they improve resumability or evidence.
5. Set the goal contract - Express done-when as verifiable outcomes (files, tests, gates, artifacts). Activate a persistent goal or stop-hook mode only when the user explicitly requests it and the tool supports it. Stop once the acceptance criteria are satisfied, verification is recorded, docs are current, known gaps are documented, and unnecessary work has been removed or deferred. Every delegated packet gets its own contract: GOAL (one sentence), DONE WHEN (verifiable), EVIDENCE (what to record), RISKS, and NEXT STEP.
6. Decompose and route - Cut work packets per docs/ai/orchestration.md and route by docs/ai/README.md. Execute serially by default. When the user explicitly requested delegation, use a read-only scout for territory mapping, bounded workers or dart-execute-packet for implementation, an independent reviewer for acceptance review, and a release auditor for branch adaptation; Codex supplies these roles as the .codex/agents/ profiles and other tools use separate sessions. Use parallel writers only with user-approved implementation delegation and explicit disjoint ownership; research/review approval alone is insufficient. Record the phase-specific mode and delegation decision per docs/ai/orchestration.md.
7. Run the autonomous work/review cycle - For each meaningful chunk: plan, execute, verify, then run an independent/specialized review lane. Treat review findings as hypotheses: investigate, fix or record no-fix evidence, clean up, re-verify, and re-review. A packet is not done until the current post-fix state has at least two clean review passes recorded.
8. Supervise and steer - Monitor progress; unblock, reassign, or re-cut packets on scope mismatch. Workers return Task, Summary, Files changed, Evidence/tests, Risks, and Recommended next step. Use another tool, an independent session, or the bounded specialist profiles within the approved model/effort and delegation scope; use role-separated local review when an independent route is unavailable under that scope. Root-cause failures and fold newly discovered unknowns back into step 3. Author role separation cannot clear publication; use the independent local gate in docs/onboarding/ai-reviews.md before any branch push.
9. Update docs at each stopping point - Follow docs/dev_tasks/README.md's Session End protocol. Keep the current snapshot sufficient for a fresh session to resume without hidden chat memory or reading the entire history.
10. Version-control and closeout - Keep commits and PRs coherent: separate feature work, bug fixes, refactors, docs, experiments, and AI-infra changes when practical; review the diff, remove unrelated changes, make the changelog decision, and run pixi run lint before commits. Run task-specific gates from docs/ai/verification.md, record evidence per packet, and complete the principle audit. A project is complete only when the north star and acceptance criteria are met, verification evidence is recorded, docs are current, known gaps are documented, unnecessary work is removed or deferred, and final state is summarized in RESUME.md or a durable owner. Promote durable artifacts out of docs/dev_tasks/<task>/ and remove the folder in the completing PR. GitHub mutations (push, PR, comments, re-triggers) only with explicit maintainer/user approval.

Prompt Shape

Use an outcome-first brief. Do not repeat this workflow's logistics or required reading in the task prompt; the capability loads them.

TASK: <one-sentence objective>

Done when:
- <verifiable outcome: a file, test, gate, benchmark, or artifact>
- <verifiable outcome>

Constraints/evidence:
- <task-specific must/never rules and owner references>
- <known risks, branch/PR facts, or required comparison>

Put this brief after /dart-ultrawork or $dart-ultrawork. When goal mode is available, make the same Done when list the goal contract.

Output

• Interview record, uncertainty-resolution evidence, and project-home path
• Packet list, routing, goal contracts, gates, and review-loop status
• Per-packet evidence, GUI/demo artifacts when relevant, and updated docs
• Principle audit, cleanup status, and approved external mutations