boi

Dispatch self-evolving specs to Claude Code workers for iterative task execution.

Updated Jul 6, 2026
One-click install
npx skills add https://github.com/mrap/boi --skill boi
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: boi
Source: https://github.com/mrap/boi/tree/main
Command: npx skills add https://github.com/mrap/boi --skill boi

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

BOI automates the decomposition and dispatch of user tasks into a self-evolving spec, executing one task at a time via Claude Code workers with fresh context each iteration.

Core Features & Use Cases

  • Self-evolving specs enable dynamic decomposition and task addition as work is discovered.
  • Phase- and pipeline-oriented execution with per-task prompts, multi-phase evaluation, and dependency management.
  • Parallelism via a fleet of Claude Code workers while keeping each iteration isolated.
  • Supports per-spec mode selection (execute, challenge, discover, generate) and runtime abstractions.

Quick Start

Dispatch a spec with boi dispatch --spec <path> to begin the self-evolving loop.

Frequently Asked Questions about boi

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

FAQPage Schema
How do I automate complex task decomposition for Claude Code workers?

You can automate task decomposition by dispatching a self-evolving spec to Claude Code workers. The system breaks down complex tasks iteratively, adding new sub-tasks dynamically as work is discovered, and executes them with fresh context each cycle.

What is a self-evolving spec in AI workflow automation?

A self-evolving spec is a Markdown document that dynamically decomposes and adds tasks as work is discovered. It uses SKILL.md frontmatter and per-task prompts to load context and guide isolated execution across parallel Claude Code workers.

How do I dispatch a spec to start an iterative execution loop?

You dispatch a spec by running the command 'boi dispatch --spec <path>'. This initiates the self-evolving loop, executing tasks one at a time via Claude Code workers with fresh context and dependency management.

Can I run parallel Claude Code workers while keeping iterations isolated?

Yes, you can run a fleet of parallel Claude Code workers while keeping each iteration isolated. The system supports phase- and pipeline-oriented execution, ensuring each worker operates with fresh context per task.

Does this approach support different execution modes for task discovery?

Yes, the system supports per-spec mode selection including execute, challenge, discover, and generate modes. These modes allow runtime abstractions for multi-phase evaluation and dependency management during task execution.

When should I use spec dispatch instead of manual task breakdown?

You should use spec dispatch for complex, multi-step tasks that benefit from self-evolving specs and parallel worker execution. If your project requires dynamic task addition and phase-oriented evaluation, this approach automates the iterative cycle effectively.