Agent Skills: dart-new-task

DART New Task: start a feature, bugfix, refactor, docs, build, or test task

ID: dartsim/dart/dart-new-task

Repository

dartsimLicense: BSD-2-Clause
1,212304

Install this agent skill to your local

npx skills add dartsim/dart --skill dart-new-task

Skill Files

The files that make up dart-new-task.

Download Skill

Loading file tree…

.agents/skills/dart-new-task/SKILL.md

Skill Metadata

Name
dart-new-task
Description
DART New Task: start a feature, bugfix, refactor, docs, build, or test task
<!-- AUTO-GENERATED FILE - DO NOT EDIT MANUALLY --> <!-- Source: .claude/commands/dart-new-task.md --> <!-- Sync script: scripts/sync_ai_commands.py --> <!-- Run `pixi run sync-ai-commands` to update -->

dart-new-task

Use this skill to run the DART dart-new-task workflow. The editable workflow source lives in .claude/commands/; this generated skill is its entrypoint for Codex and Muse.

Invocation

  • Claude Code: /dart-new-task <arguments>
  • Codex: $dart-new-task <arguments>
  • Muse: /dart-new-task <arguments>

Treat the text after the skill name, or the user's request when the skill is loaded on demand, as $ARGUMENTS. When the workflow references $1, $2, etc., map those to the positional values supplied by the user.

Command Body

Start a new task in DART: $ARGUMENTS

Required Reading

Read these files first: @AGENTS.md @docs/onboarding/building.md @docs/onboarding/contributing.md @docs/onboarding/release-management.md @docs/onboarding/code-style.md @docs/dev_tasks/README.md @docs/ai/sessions.md @docs/ai/principles.md @docs/ai/verification.md

Workflow

  1. Understand the task - Parse: goal, constraints, type (feature|bugfix|refactor|docs)
  2. Assess scope - Multi-phase or multi-session? Create docs/dev_tasks/<task>/ (see docs/dev_tasks/README.md for criteria). Team-scale work (multiple parallel lanes needing orchestrated worker agents) switches to dart-ultrawork instead. For multi-session, design-heavy, public API, solver/paper, release, or cross-module work, fill the dev-task specification intake before editing: value, scope, assumptions, traceability, non-goals, acceptance evidence, gates, and open decisions. If consequential ambiguity would change public API, release compatibility, numerical correctness, benchmark claims, or roadmap scope, record an owner-local Decision needed block instead of silently choosing.
  3. Setup - Resolve the target from the live PR base, or use main for new development and fixes. Release-specific packaging, CI, branch guidance, and backports use the selected maintained release-* branch; backports of merged main fixes use dart-backport-pr. Fetch the target and read its docs/onboarding/release-management.md § "Release target" before creating a topic branch. Verify the exact DART <Next release> milestone is open on GitHub. Publishing requires explicit maintainer/user approval. Create the topic branch without tracking its base: git switch --no-track -c <type>/<topic> origin/<target-branch>.
  4. Implement - Keep commits focused, follow code style. Delegate by default under docs/ai/orchestration.md, honoring a per-task delegation=off opt-out: route bounded packets to the best available agent, and keep small tasks or a single hard workstream with one agent when proportionate.
  5. Verify - Run pixi run lint before committing, then run the focused branch gate for the touched surface. Use pixi run test-all when feasible, and pixi run -e gazebo test-gz for package, collision, constraint, or downstream Gazebo/gz-physics compatibility work. If the claim depends on scene structure, simulation, dynamics, collision/contact, Filament output, or a visual example, route through dart-verify-sim: prove correctness with text first, then add assessed claim-tied visual/debug evidence or record why it is unavailable or not applicable. Then run the review loop in docs/ai/orchestration.md § "Review Loop": at least two clean independent review passes on the post-fix state, preferably from a different client or model family than the writer (under delegation=off, a role-separated local review).
  6. PR - After explicit maintainer/user approval, push with the same local and remote topic-branch name: branch=$(git branch --show-current); git push -u origin "HEAD:${branch}". Then create the draft PR against <target-branch> using .github/PULL_REQUEST_TEMPLATE.md and the exact open GitHub milestone DART <Next release> resolved from the fetched base's release-target table.
  7. Cleanup - Before PR: if task used docs/dev_tasks/<task>/, first promote durable dashboards, evidence matrices, API inventories, migration maps, or long-lived decisions into release/onboarding/AI docs. Then remove the dev-task folder completely (include the deletion in this PR, not after merge).

Type-Specific

  • Bugfix: Target the development branch (main)
  • Refactor: No behavior changes
  • Feature: Add tests + docs
  • Scope: Follow the resolved branch's AGENTS.md compatibility and approval rules: on main that includes approved PLAN-624 work; release branches take compatible maintenance, dependency minimization, CI, docs, and fixes.

Output

  • Task type, branch, and files changed
  • Verification commands run and their results
  • Remaining steps and anything held for explicit maintainer/user approval before push/PR

More agent skills