Building LockIn

github

Naman Khandelwal • Peerlist

<h3>LockIn: The open-source notebook your AI tools write to.</h3>
LockIn is an open-source notebook designed to capture and organize the knowledge generated by your AI coding agents. It acts as a persistent memory for tools like Claude Code, Codex, and Cursor, ensuring that valuable insights and notes are not lost when a session ends.
<ul>
<li>Persistent Memory for AI Agents: Save notes generated by your AI coding assistants directly into your notebook.</li>
<li>Organized Knowledge: File notes under specific subjects and easily retrieve them later.</li>
<li>Code Context Awareness: LockIn tracks which code commit a note was relevant to, ensuring agents refer to up-to-date information. It checks code changes and confirms notes, replacing outdated ones or flagging discrepancies.</li>
<li>Disagreement Resolution: When notes conflict, LockIn displays their commit dates, allowing you to identify and use the most recent information.</li>
<li>Task Handoff: Seamlessly hand off mid-task to another agent or session, preserving the goal, state, and next steps.</li>
<li>Browser-Based and Private: Run LockIn in your browser using WebGPU for privacy, or self-host it on your own machine with Docker. Your model keys remain in your browser.</li>
<li>Search and Ask: Chat with any model about your notes, with answers grounded in your saved information and sources clearly cited.</li>
<li>Exportable Notes: Your notes are exportable in Markdown and JSON formats.</li>
<li>Self-Hosting: Easily set up your own instance using Docker, with support for Postgres, PDF text extraction, and server-side search.</li>
</ul>
LockIn is open-source under the AGPL-3.0 license, promoting transparency and community contribution.