Promemo Documentation

Promemo is Git-native project memory for AI-assisted development.

Git stores what changed. Promemo stores why it was built this way.

Promemo keeps durable project memory as Markdown inside your repository under .promemo/. Architecture notes, decisions, APIs, data models, TODOs, handoff context, and assistant prompts become reviewable project artifacts that can be diffed, merged, and committed like code.

Promemo is local-first:

• No hosted backend is required.
• Markdown is the durable source of truth.
• Generated memory is stored in the repository.
• Raw transcripts are not stored.
• CLI and MCP integrations use the same core engine.
• AI assistants can read and write memory without owning the memory format.

Why Promemo

AI coding tools are useful, but most of them forget the project between sessions. The result is repeated explanations, stale assumptions, missing context, and architecture decisions that live only in chat history.

Promemo gives projects a memory layer that is:

• Git-native: memory lives beside code.
• Human-readable: Markdown files can be opened and reviewed directly.
• Assistant-friendly: JSON contracts and MCP tools make memory usable from automation and AI clients.
• Reviewable: generated files can be inspected before they are committed.
• Local-first: no server is needed to keep useful project memory.

Use Promemo when you want an assistant to remember:

• why a feature was implemented a certain way
• which files matter for a workflow
• what API or data model decisions were made
• what TODOs or open questions remain
• what context should be loaded before future work
• what changed in a work session before writing a handoff

Next Steps

• Learn the core concepts.
• Install Promemo and run the quick start.
• Connect Promemo to an assistant with MCP integration.