Show HN: Memory Lane – bootstrap your naive Claude instances with their history

It seems like a lot of us work in either "short context" (ie, tell the LLM to do a think real quick and then end the instance) practice or "long context" (big project, lot of brainstorming across context heaps).

If you, like me, prefer the latter, you might like this tool.

It watches the JSONL log files that claude code produces and saves them in a django app. You can browse them via a frontend, and your agent can access them through the MCP server.

When you start a new claude code instance, it can call the boostrap MCP function to "become itself" again. I can't imagine working with claude without it anymore.