Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

start-work (Plan Execution Orchestrator)

An orchestrator skill that executes an existing work plan by fanning it out to parallel subagents and enforcing evidence-backed verification before any checkbox is closed.

AutomationAdvanced★ 67,531⑂ 5,507AI score 7/10Last updated: Aug 9, 2026

What it does

start-work takes an already-written plan under .omo/plans/*.md and drives it to completion as a strict orchestrator: the root agent never edits product files — every implementation, test, and QA task is delegated to a spawned worker.

  • Selects the plan, writes Boulder state (.omo/boulder.json), then works checkbox by checkbox
  • Dispatches all independent sub-tasks in one parallel spawn burst; only named dependencies serialize
  • A delegation router maps work shape to executor tier (quick / visual-engineering / writing / deep / ultrabrain)
  • Five mandatory gates per checkbox: plan reread, automated verification, real manual-QA artifact, 9 adversarial QA classes, cleanup receipts
  • Appends every result to .omo/start-work/ledger.jsonl as JSONL evidence
  • --make-pr / --ship create a task-owned git worktree, open a PR, chase CI and review, then merge

Who it's for

  • Teams that split planning and execution into separate phases
  • Anyone tired of "tests pass, therefore done" claims without real proof
  • Work that spans many files and benefits from parallel subagents
  • Developers who want the loop automated all the way to a merged PR

Examples

  1. Resume a plan: say "start work" and it finds the next unchecked box in .omo/plans/refactor-auth.md, runs it, verifies, then flips - [ ] to - [x].
  2. PR mode: $start-work payment-api --make-pr creates a dedicated worktree, gives each independent lane its own worker, then opens a reviewer-ready PR and hands back the URL.
  3. Full delivery: $start-work --ship keeps going after the PR opens — routing CI failures and review comments back to workers, merging, then removing the worktree.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the packages/shared-skills/skills/start-work folder from the GitHub repo code-yeongyu/oh-my-openagent into my ~/.claude/skills/start-work/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/code-yeongyu/oh-my-openagent /tmp/oh-my-openagent && mkdir -p ~/.claude/skills && cp -r /tmp/oh-my-openagent/packages/shared-skills/skills/start-work ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal and clone the repository: git clone https://github.com/code-yeongyu/oh-my-openagent
  2. Create the skills folder if needed: mkdir -p ~/.claude/skills
  3. Copy just this skill: cp -r oh-my-openagent/packages/shared-skills/skills/start-work ~/.claude/skills/
  4. Confirm that ~/.claude/skills/start-work/SKILL.md exists.
  5. Restart Claude Code, then type "start work" or "$start-work" to check that the skill triggers.
  6. The skill expects a plan file under .omo/plans/. If you don't have one, install the companion planning skill (ulw-plan) the same way.
  7. Before the first run, make sure the project allows writes to a .omo/ directory and that git worktree is available.
View source on GitHub ↗License: NOASSERTION