blueprint

Transform high-level objectives into structured multi-agent construction plans.

3|2|Updated Mar 8, 2026

Search Tags:#workflow #planning #multi-agent #blueprint #software-engineering #project-management #adversarial-review

Ready to run

npx skills add https://github.com/agentmatters/mullai-bot --skill blueprint-agentmatters

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: blueprint
Source: https://github.com/agentmatters/mullai-bot/tree/main/src/Mullai.Skills/Skills/claude-code-everything/blueprint
Command: npx skills add https://github.com/agentmatters/mullai-bot --skill blueprint-agentmatters

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Turn complex objectives into step-by-step, reusable plans for multi-session, multi-agent engineering tasks, ensuring context is preserved between steps.

Core Features & Use Cases

Adversarial review gate to validate plans before execution.
Dependency graph, parallelism identification, and rollback strategies.
Cold-start, self-contained steps that can be executed by a fresh agent without prior context.
Use Cases: Planning a large refactor that spans multiple PRs and teams.

Quick Start

Provide a clear objective and let Blueprint generate a detailed, multi-session construction plan.

Frequently Asked Questions about blueprint

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

FAQPage Schema

How do I turn a complex software engineering objective into a multi-step plan?▼

A multi-agent construction plan creates self-contained steps that a fresh agent can execute without prior context, preserving context between steps and identifying parallelism for coordinated work across sub-agents.

How do I plan a large refactor spanning multiple pull requests?▼

Adversarial review in project planning validates the generated construction plan before execution, ensuring the self-contained steps, dependency graph, and rollback strategies are robust enough for coordinated multi-agent engineering tasks.

What is the best way to coordinate multi-agent engineering tasks without losing context?▼

Blueprint applies a five-phase process—research, design, draft, review, and register—to transform a high-level objective into a structured, step-by-step construction plan suitable for multi-agent engineering projects.

Does the planning workflow support parallelism identification for sub-agents?▼

The five-phase planning process requires an adversarial review gate to validate the multi-session plan before execution, ensuring dependency graphs and rollback strategies are sound for multi-agent engineering projects.