ADHD Progress Hub

ADHD Progress Hub is an open-source AI developer tool that preserves coding-session continuity so AI assistants like Claude Code, Cursor, and Codex can resume context instead of starting cold. It keeps a record of unfinished work as named threads, each holding progress notes and concrete next steps, so a developer who switches tools or returns after a break can pick up exactly where they left off. Before a new task begins, its overlap detection checks existing threads for duplicate effort, which matters when the same project is touched from several AI coding environments. The hub exposes three interfaces from one store: the MCP protocol for AI assistants to read and write progress directly, a REST API, and a web dashboard for humans. Work is held in a SQLite database alongside Markdown progress files, and the project recommends Docker Compose for a persistent self-hosted server, with an optional local subprocess mode for single-machine setups. Optional integrations include GitHub or Gitea sync, OpenClaw notifications, and local transcript indexing, plus reminders on once, per-session, daily, or random schedules to resurface stale threads. Because it is MIT-licensed and self-hosted, teams keep all session memory on their own infrastructure. For developers evaluating the best AI tool for cross-session memory, ADHD Progress Hub is a self-hosted option that treats progress as shared state across every assistant that connects.

Keeping cross-session memory for AI coding assistants: tracking unfinished threads, next steps, and detecting duplicate work across Claude Code, Cursor, and Codex.