Mnemos
@arhuman
About Mnemos
Local-first MCP memory server for AI agents, with source citations and OKF/Markdown knowledge bases.
Config
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"mnemos": {
"command": "mnemos",
"args": [
"serve",
"--mnemos-dir",
"/abs/path/to/.mnemos"
]
}
}
}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 Mnemos?
Mnemos is a local memory server for AI agents that indexes plain-Markdown files and serves them over the Model Context Protocol (MCP), so agents like Claude Code can search, read, and cite your own knowledge instead of guessing.
How to use Mnemos?
Install by cloning the repository and running make install (requires Go 1.25+). Index a project with mnemos init and mnemos ingest docs --collection myproject, then wire it into Claude Code with claude mcp add mnemos -- mnemos serve --config /abs/path/to/project/.mnemos.toml.
Key features of Mnemos
- Truly local-first: runs entirely on your machine with no network or telemetry.
- Zero dependencies: one self-contained, cgo-free Go binary.
- Works with any MCP client, built for Claude Code.
- Cited answers: every result links to exact file, section, and line range.
- Fast search via SQLite FTS5/bm25; optional local semantic and hybrid search.
- Read-write memory: agents can capture notes, list, forget, and move entries.
- Safe by default: read-only unless opted in; writes are path-confined and secret-scanned.
Use cases of Mnemos
- Give Claude Code cited access to your ADRs and design documents.
- Index project notes, runbooks, and source code for instant retrieval.
- Create a durable project memory that agents can both read and write.
- Consolidate raw captures into canonical docs with automatic conflict handling.
FAQ from Mnemos
How does Mnemos handle document identity?
A document’s identity is its path relative to the scan root you ingested (e.g., docs). If two ingested directories each contain a file with the same relative path, the second ingest silently overwrites the first.
Why does the --config path have to be absolute?
Claude Code does not guarantee the working directory it spawns the server in. An absolute --config path ensures that the database, capture directory, and tree root anchor to that file regardless of where the server is launched.
Does Mnemos require a vector database or external services?
No. Mnemos uses SQLite FTS5 and BM25 for default lexical search. No vector database, Ollama, Python, or Node service is needed.
What are Mnemos’s transport and authentication mechanisms?
Mnemos is built for MCP and runs as a local server on your machine. Authentication is not mentioned; it is intended for local, trusted environments.
What is the difference between the default lexical build and the optional semantic build?
The default lexical build uses SQLite FTS5/BM25 and works well for keyword-based queries. The optional semantic build (enabled with a build tag) uses an on-device embedding model for natural-language and hybrid search, improving retrieval on queries that share no keywords with the target documents.
Frequently asked questions
How does Mnemos handle document identity?
A document’s identity is its path relative to the scan root you ingested (e.g., `docs`). If two ingested directories each contain a file with the same relative path, the second ingest silently overwrites the first.
Why does the `--config` path have to be absolute?
Claude Code does not guarantee the working directory it spawns the server in. An absolute `--config` path ensures that the database, capture directory, and tree root anchor to that file regardless of where the server is launched.
Does Mnemos require a vector database or external services?
No. Mnemos uses SQLite FTS5 and BM25 for default lexical search. No vector database, Ollama, Python, or Node service is needed.
What are Mnemos’s transport and authentication mechanisms?
Mnemos is built for MCP and runs as a local server on your machine. Authentication is not mentioned; it is intended for local, trusted environments.
What is the difference between the default lexical build and the optional semantic build?
The default lexical build uses SQLite FTS5/BM25 and works well for keyword-based queries. The optional semantic build (enabled with a build tag) uses an on-device embedding model for natural-language and hybrid search, improving retrieval on queries that share no keywords with the target documents.
Basic information
More Other MCP servers

DeepMark
DeepMark helps teachers deliver rapid, consistent marking with meaningful feedback for every student — in a fraction of the time. What once took a week, now takes one free period.
Activepieces
activepiecesAI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents
Mobile Mcp
mobile-nextModel Context Protocol Server for Mobile Automation and Scraping (iOS, Android, Emulators, Simulators and Real Devices)
Inbox Zero AI
elie222The world's best AI personal assistant for email. Open source app to help you reach inbox zero fast.

Peekaboo MCP – lightning-fast macOS screenshots for AI agents
steipetePeekaboo is a macOS CLI & optional MCP server that enables AI agents to capture screenshots of applications, or the entire system, with optional visual question answering through local or remote AI models.
Comments