IdlerGear

Active Development

A knowledge management system for AI-assisted development sessions. Captures tasks, notes, decisions, and context across sessions so AI coding assistants can maintain continuity and never lose track of work in progress.

🔗 GitHub Repository

Overview

AI coding assistants are powerful but stateless—each conversation starts fresh with no memory of previous work. IdlerGear solves this by providing persistent knowledge management that survives across sessions, enabling true continuity in AI-assisted development.

Features

  • Session continuity - Save and restore working context between AI sessions
  • Task management - Track work items with GitHub Issues integration
  • Note capture - Document discoveries, gotchas, and decisions as you work
  • Project vision - Maintain high-level goals and architecture principles
  • Multi-agent coordination - Enable multiple AI assistants to work together without conflicts
  • Git integration - Link commits to tasks, track work across branches
  • MCP server - Exposes tools via Model Context Protocol for Claude Code and other AI assistants

Use Cases

  • Long-running projects - Maintain context across weeks or months of development
  • Knowledge capture - Document "why" decisions for future reference
  • Team coordination - Multiple developers (human or AI) working on the same codebase
  • Context preservation - Never lose track of TODO items or half-finished work

Architecture

IdlerGear operates as a daemon that AI assistants interact with via an MCP (Model Context Protocol) server:

[AI Assistant] ←→ [IdlerGear MCP Server] ←→ [IdlerGear Daemon]
                                                    ↓
                                         [Knowledge Store]
                                         - Tasks
                                         - Notes
                                         - Vision
                                         - Session State

Knowledge is stored locally in .idlergear/ directory, with optional GitHub backend for tasks.

Philosophy

FORBIDDEN: TODO.md, NOTES.md, SESSION_*.md files, // TODO: comments

USE INSTEAD: idlergear task create, idlergear note create, idlergear vision edit

By forcing structured knowledge capture through tools rather than free-form markdown files, IdlerGear ensures that context is preserved in a format AI assistants can reliably query and act upon.