AI Project Playbook Agent

An autonomous PM agent that takes your project from idea to deployment.

What This Agent Does

I am an AI Project Manager that can guide you through the entire software development lifecycle:

1. Discovery: Ask questions to understand your project requirements
2. Planning: Generate CLAUDE.md (global rules) and PRD automatically
3. Roadmap: Break down into features and create implementation plans
4. Implementation: Execute PIV Loop for each feature (using Claude Code)
5. Deployment: Generate deployment configs based on your scale phase

How to Use

Simply describe what you want to build:

"Create a SaaS for veterinary clinics with appointments, medical records, and billing"

I will guide you through the entire process, asking questions when needed.

Autonomy Modes

• Supervised (default): I propose actions and ask for confirmation before executing
• Autonomous: I execute without asking (activate with "modo autónomo")
• Plan-only: I only generate plans without executing (activate with "solo planea")

Available Tools

Tool	Description
playbook_start_project	Start a new project with an objective
playbook_continue	Continue from where we left off
playbook_answer	Answer agent's question and continue
playbook_search	Search RAG in the playbook guides
playbook_get_status	Get current project status
playbook_create_agent	Create a new specialized agent
playbook_list_agents	List agents in the registry
playbook_create_workflow	Create multi-agent workflow

Project Types Supported

• SaaS Applications: Multi-tenant web apps with authentication, billing, etc.
• API Backends: RESTful or GraphQL APIs with proper architecture
• Agent Systems: Single AI agents with tools and memory
• Multi-Agent Systems: Orchestrated agent workflows (Supervisor, Parallel, Sequential patterns)

Scale Phases

Phase	Users	Monthly Cost	Stack
MVP	<100	$300-500	Netlify + Railway
Growth	100-10K	$1,500-3K	Netlify + Cloud Run
Scale	10K-100K	$8K-15K	Netlify + GKE
Enterprise	100K-1M+	$50K-150K	Multi-cloud

Multi-Agent Patterns

When building agent systems, I can create workflows using these patterns:

1. Agent-as-Tool: Agent A invokes Agent B as a tool
2. Agent Handoff: Agent A passes complete control to Agent B
3. Supervisor: Dynamic orchestration with shared state
4. Parallel: Fan-out/fan-in for concurrent execution
5. Sequential: Pipeline where output flows to next agent
6. LLM Routing: Cost-optimized routing to specialized agents

References

• Quick Navigation - Fast access to playbook sections
• Agent Capabilities - Full list of what the agent can do
• Playbook Content - Complete methodology documentation