# junto-memory MCP server

> Provides persistent shared memory and multi-agent coordination for AI coding agents using MongoDB and ChromaDB.

- Repository: https://github.com/tlemmons/junto-memory
- Page: https://hostedmcp.org/servers/tlemmons/junto-memory
- Transport: —
- Stars: 6
- License: MIT

## Tools (35)

- `memory_start_session`
- `memory_end_session`
- `memory_query`
- `memory_get_by_id`
- `memory_store`
- `memory_record_learning`
- `memory_search_global`
- `memory_list_projects`
- `memory_lock_files`
- `memory_unlock_files`
- `memory_get_locks`
- `memory_add_backlog_item`
- `memory_list_backlog`
- `memory_update_backlog_item`
- `memory_complete_backlog_item`
- `memory_send_message`
- `memory_notify`
- `memory_get_messages`
- `memory_acknowledge_message`
- `memory_update_message_status`
- `memory_heartbeat`
- `memory_get_agent_status`
- `memory_list_agents`
- `memory_register_function`
- `memory_find_function`
- `memory_enrich_function`
- `memory_get_enrichment_queue`
- `memory_become_librarian`
- `memory_define_spec`
- `memory_get_spec`
- `memory_list_specs`
- `memory_project`
- `memory_checklist`
- `memory_db`
- `memory_admin`

## FAQ

### How should I install this server?

The server is designed to be installed by an AI agent rather than a human. You should clone the repository and provide the installation prompt to an MCP-capable agent like Claude Code.

### Does it support local models?

Yes, agents can delegate routine text tasks like summarization or classification to a local model running via Ollama.

### Can I use it with just one agent?

Yes, a single agent can benefit from persistent memory, function registries, and task backlogs even without multi-agent coordination features.