Rekindle MCP server

Rekindle is a small helper that gives your AI a memory between chats. It saves a short checkpoint at the end of a session and reads it back at the start of the next one, so you spend less time re-explaining your project. It is handy if you work with Claude Code or Cursor on the same project day after day.

What this MCP server does

When a session starts, Rekindle puts together a short orientation report: who you are working with, what was decided last time, where you left off, and what still needs follow-up. Your AI reads that report and starts work already knowing the context. During a long session, if the chat gets trimmed down, Rekindle can save what would otherwise be lost. At the end, your AI writes a structured close-out: the checkpoint, decisions, open questions, and things to watch for next time. Everything lives in one local file, and nothing is sent anywhere.

Click to zoom

What you can do with it

• Start a session with a short report of where you left off
• Save a checkpoint, decisions, and open questions at the end of a session
• Search your past notes by keyword, ranked by how important they are
• Browse and clean up stored memories by category or project
• Capture the current session on demand before the chat gets trimmed
• Read back an earlier capture as a short summary or in full
• Flag what is missing at the start of a session, like an empty profile or stale notes

What it gives back to you

You get short written answers in the chat: a boot report with a score, a list of past memories with their categories, or a summary of an earlier capture. When your AI saves a session, it confirms what was stored, like the checkpoint and the open questions. Searches come back as a ranked list of matching notes. Everything is text you can read right in the conversation.

Before you start

Who it's for

People who work on the same project with an AI assistant over many days, like developers, writers, and anyone tired of repeating themselves.