start-work

Orchestrate Prometheus work plans with subagents and Boulder state updates.

17|Updated May 29, 2026
One-click install
npx skills add https://github.com/code-yeongyu/lazyclaudecode --skill start-work-code-yeongyu
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: start-work
Source: https://github.com/code-yeongyu/lazyclaudecode/tree/main/plugins/omo/skills/start-work
Command: npx skills add https://github.com/code-yeongyu/lazyclaudecode --skill start-work-code-yeongyu

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates the end-to-end execution of Prometheus work plans within Claude Code, handling Boulder state, ledger updates, and subagent coordination so users can transition from planning to active work with minimal friction.

Core Features & Use Cases

  • Phase-driven plan execution with Boulder state management and session continuation.
  • Parallel subagent orchestration and Stop-hook continuation to support long-running workflows.
  • Worktree-aware planning, evidence ledger integration, and automatic plan advancement while preserving reproducibility.

Quick Start

Start a work plan by invoking the start-work command with your plan name and an optional --worktree path.

Frequently Asked Questions about start-work

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I execute a Prometheus work plan inside Claude Code?

To execute a Prometheus work plan in Claude Code, invoke the start-work command with your plan name. It orchestrates the end-to-end execution from planning to completion while coordinating parallel subagents.

What is Boulder state management and how does it track plan progression?

Boulder state management tracks plan progression by updating a state file located at .omo/boulder.json. It enforces safe, auditable advancement with mandatory checks and ledger entries to preserve reproducibility.

Can I run parallel subagents and maintain an evidence ledger for long-running workflows?

Yes, you can run parallel subagents and maintain an evidence ledger. The execution supports Stop-hook continuation to handle long-running workflows and ensures all plan updates are auditable.

Do I need a dedicated git worktree to start a work plan?

A dedicated git worktree is optional for starting a work plan. You can provide a worktree path using the --worktree flag to isolate worktree-aware planning, but it is not strictly required.

How does the session-scoped codex prefix enforce workflow rules during plan execution?

The session-scoped codex prefix enforces workflow rules by mandating prefixed sessions during plan execution. This ensures safe, auditable plan progression with mandatory checks and continuous ledger entries.

Are there limitations when using start-work for plan orchestration?

Plan orchestration is limited to plans stored under the .omo/plans directory. It requires strict adherence to mandatory checks and ledger entries, meaning any workflow bypassing these rules cannot be safely advanced.