EvolutionDB Long Term Memory
Give Claude Desktop or Claude Code durable, queryable long-term memory by pointing an stdio MCP server at a local EvolutionDB instance over the PostgreSQL wire protocol.
Overview
EvolutionDB Long-Term Memory is a MCP server for the Build phase that persists agent memory through EvolutionDB via a Python stdio bridge for Claude Desktop and Claude Code.
What is this MCP server?
- stdio MCP server (PyPI mcp-server-evolutiondb v1.9.2) for Claude Desktop and Claude Code
- Backed by EvolutionDB reachable via EVOSQL_HOST, EVOSQL_PORT (default 5433), user, password, and database env vars
- Persistent long-term memory semantics instead of single-chat context window limits
- Python3 runtime with local database control for indie builders who self-host
- Fits self-hosted stacks where you already run or can run EvolutionDB beside your agent
- Package version 1.9.2 on PyPI identifier mcp-server-evolutiondb
- Default EvolutionDB wire port 5433 documented in server.schema environmentVariables
- stdio transport with python3 runtimeHint
Community signal: 5 GitHub stars.
What problem does it solve?
Agent sessions forget prior decisions and project facts, forcing solo builders to re-explain context every time they open Claude Desktop or Claude Code.
Who is it for?
Self-hosted Claude users who want local, database-backed long-term memory with full control over host, port, and credentials.
Skip if: Builders who refuse to run EvolutionDB locally, need a zero-setup hosted memory SaaS only, or want finance or SEO tooling instead of persistence.
What do I get? / Deliverables
After configuring EVOSQL variables and the stdio server, your agent can read and write durable memory stored in EvolutionDB across sessions.
- Cross-session memory backed by EvolutionDB instead of ephemeral chat context
- Local stdio MCP integration for Claude Desktop and Claude Code
- Operator-controlled datastore with standard connection env vars
Recommended MCP Servers
Journey fit
Persistent agent memory is installed while you are building and hardening the agent workflow—sessions, tools, and context policies—before you treat production monitoring as the main concern. Agent-tooling is the canonical shelf because EvolutionDB Memory is packaged as mcp-server-evolutiondb for coding agents, not as a customer-facing analytics or SEO surface.
How it compares
Self-hosted database memory MCP over EvolutionDB, not a remote whale-analytics or knowledge-graph SaaS.
Common Questions / FAQ
Who is EvolutionDB Long Term Memory for?
It is for solo builders using Claude Desktop or Claude Code who self-host EvolutionDB and want MCP-accessible persistent memory.
When should I use EvolutionDB Long Term Memory?
Use it during Build agent-tooling when you are wiring MCP servers and need memory to survive across coding sessions on your own machine or VPS.
How do I add EvolutionDB Long Term Memory to my agent?
Install mcp-server-evolutiondb from PyPI with python3, start EvolutionDB on port 5433 (or your EVOSQL_PORT), set EVOSQL_HOST, EVOSQL_USER, EVOSQL_PASSWORD, and EVOSQL_DATABASE, then add the stdio server to your Claude MCP configuration.