Memory Arbiter Mcp
@billy12151
About Memory Arbiter Mcp
Share one memory store across all your AI coding tools. Memory Arbiter is a lightweight, fully local MCP Server with built-in dual-timeline conflict arbitration. No cloud, no LLM calls — just SQLite.
Basic information
Config
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"memory-arbiter": {
"command": "memory-arbiter-mcp",
"env": {
"MEMORY_ARBITER_CLIENT": "zcode",
"MEMORY_ARBITER_AGENT_ID": "zcode-default",
"MEMORY_ARBITER_DB_PATH": "~/.local/share/memory-arbiter/memory.sqlite3"
}
}
}
}Tools
No tools detected
We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.
Overview
What is Memory Arbiter MCP?
Memory Arbiter MCP is a Model Context Protocol server that provides a shared local memory store, backed by SQLite, for multiple AI coding tools such as ZCode, Codex, Cursor, and Claude Code. It is designed for developers who need a consistent memory layer across different assistants without relying on cloud services.
How to use Memory Arbiter MCP?
Install the package with pip install memory-arbiter-mcp, then add it as an MCP server to each AI coding tool you use. The shared memory database is stored locally in an SQLite file; no additional infrastructure is required.
Key features of Memory Arbiter MCP
- Shared memory store across multiple AI coding tools
- Conflict arbitration with multi-factor resolution
- User-confirmed memories locked from agent overwrites
- Zero cloud dependencies, all local SQLite
- Graceful degradation through fallback search methods
Use cases of Memory Arbiter MCP
- Maintain a consistent knowledge base across different coding assistants
- Preserve user-confirmed context that agents cannot accidentally overwrite
- Resolve conflicting memory entries automatically during multi-agent workflows
- Keep all memory data locally for privacy and offline usage
- Fall back to simpler search methods when vector search is unavailable
FAQ from Memory Arbiter MCP
How does conflict arbitration work?
Conflicts are resolved using a dual-timeline priority: user-confirmed entries take precedence, then event time, source trust level, and finally ingest time.
Can agents overwrite memories I’ve confirmed?
No. User-confirmed memories are locked and cannot be overwritten by agents, ensuring critical context remains stable.
Does this server require cloud services or LLM calls?
No. Everything runs locally on SQLite with no cloud dependencies and zero LLM calls for memory operations.
What happens if sqlite-vec is not installed?
The server degrades gracefully through fallbacks: first FTS5, then LIKE queries, and finally reading from a JSONL backup.
How is the memory store shared between tools?
Each tool connects to the same MCP server, which reads and writes a single local SQLite database file.
More Memory & Knowledge MCP servers
Obsidian MCP Server
cyanheadsRead, write, search, and surgically edit Obsidian vault notes, tags, and frontmatter via MCP. STDIO or Streamable HTTP.
JupyterMCP - Jupyter Notebook Model Context Protocol Integration
jjsantos01A Model Context Protocol (MCP) for Jupyter Notebook
Notion MCP Server
makenotionOfficial Notion MCP Server
Solomd
zhitongblogA markdown editor — and the bridge to your LLM. Local-first, MIT, ~15 MB. Bundled MCP server lets Claude Code / Codex / Cursor drive your vault directly. 14 AI providers BYOK.
MemoryMesh
CheMiguel23A knowledge graph server that uses the Model Context Protocol (MCP) to provide structured memory persistence for AI models.
Comments