Agentic Orchestration Toolkit

While developing the Mandy RAG engine, I needed a way to coordinate multiple AI agents and manage complex development tasks without losing context. This led to the creation of a suite of specialized CLI tools designed for an "Agent-Native" workflow.

🤖 Squadron: Multi-Agent CLI Orchestrator

A task management system that allows running complex coding tasks across different AI agents from a single interface.

• Provider Agnostic: Swap between local and cloud models based on the task complexity.
• Context Isolation: Ensures agents only see the relevant parts of the codebase.
• Auto-Approve Mode: "YOLO" execution for rapid, iterative development.

View on GitHub

🗂 Holoscanner: Git-Backed PM

A project management tool where every "sticky note" is a version-controlled markdown file.

• Git as a Database: Every change creates a commit, providing a perfect audit trail.
• Markdown-First: Human-readable tasks that live right alongside your code.
• Async-First: Designed for seamless collaboration between human architects and AI agents.

View on GitHub

📄 pym-particle: Document-to-Markdown Pipeline

A high-performance pipeline for converting complex documents (PDF, DOCX, etc.) into clean Markdown for RAG ingestion.

• Structure Preservation: Detects headings, tables, and lists without summarizing.
• LLM-Enhanced: Optional verification pass for cleaning up complex layouts.

View on GitHub

---

Status: Open Source