myICOR/myPKA

What it solves

myPKA is a local, AI-powered personal knowledge management (PKM) system that eliminates vendor lock-in and data fragmentation. It replaces the need for separate note-taking, planner, and health apps by providing a structured folder architecture that an AI team can manage, ensuring your data remains in plain markdown on your own machine.

How it works

The system operates as a scaffold of folders and markdown files that you point any LLM (such as Claude, Gemini, or Cursor) toward. It implements a multi-agent architecture where a lead orchestrator (Larry) routes requests to a team of six specialized agents (Nolan, Pax, Penn, Mack, and Silas) who handle tasks like deep research, filing notes, automation, and database architecture. The team uses a task-tracking system and individual journals to maintain continuity across sessions, ensuring that unfinished work is picked up where it left off.

Who it’s for

• Knowledge workers seeking a transparent, local alternative to SaaS tools like Notion or Tana.
• Founders and operators managing multiple complex projects that require cross-linking across people, goals, and topics.
• Generalists and parents needing a structured way to to organize diverse inputs (health, school, ideas).
• AI tinkerers looking for a production-ready reference architecture for multi-agent systems.

Highlights

• LLM-Agnostic: Works with any LLM tool that can read and write to a local directory.
• Plain Markdown: All data is stored as .md files, making it fully compatible with Obsidian and other text editors.
• Multi-Agent Team: Includes six pre-loaded specialists with defined contracts and SOPs.
• myPKA Cockpit: A free, local, wikilink-aware interface for navigating the scaffold.
• Zero Lock-in: Supports importing from Notion, Heptabase, Roam, and other PKM tools without proprietary formats.
• Scalable Storage: Can upgrade from plain files to a SQLite mirror for faster lookups as the knowledge base grows.