memory-lane
AI coding agents that remember you tomorrow.
Why it exists
Every AI coding session starts from zero. The agent forgets your preferences, your project conventions, and the decisions you made together yesterday - so you spend the first ten minutes of every session re-explaining yourself.
We wanted memory that persists across sessions and across tools, without shipping our working context to someone else's cloud.
What it does
- Saves durable memories - preferences, decisions, project facts - and recalls the relevant ones automatically at the start of each session.
- Semantic retrieval surfaces only what matters for the task at hand instead of dumping everything into context.
- An approval workflow keeps memory curated: the agent suggests, you approve, nothing sneaks in.
- A CLI and MCP integration make it harness-agnostic - it works with Claude Code and any agent runtime that can shell out or speak MCP.
Under the hood
- TypeScript, local-first by design: your memories live on your machine, not in a hosted database.
- Memories are plain, inspectable records - you can read, edit, and delete everything the agent knows.
- Built for the daily loop: save in one session, recall in the next, with continuity notices when workflow agreements change.
Why it matters
This is context engineering - the discipline behind every serious AI product we build for clients. If your business needs an assistant that remembers your customers, your policies, and your history, this is the same craft applied.