memnest

An MCP memory service that lets multiple AI coding tools search the same local memory

Visit website ↗

SITE CAPTURE

Based on information published on the service's website. Selection and review policy

WHAT IT DOES

What does it do?

memnest uses a single Rust service and SQLite to store memories and conversation history locally and make them available to MCP and HTTP clients. Embeddings and hybrid search run locally without calling an LLM. It does not automatically import built-in memories, and all clients must connect to the same service.

KEY FEATURES

Key features

Memory search across sessions and tools
Local SQLite storage
Local embeddings and hybrid search
Streamable HTTP MCP connection
JSON HTTP API and Stdio MCP
Automatic workspace-scoped context

Developer tools
AI
Open source

Report this service

GROWTH SIGNAL

Version history

We revisit public service pages about every four weeks and record meaningful changes to functionality or direction.

VERSION 1August 2026
First recorded
First featured on Raiders.
VERSION 2September 2026
Clarified how to verify cross-tool memory and its limitations
A concrete cross-client example and curl demo were added, showing how to retrieve a decision saved in Claude Code in a later pi or Codex session. The page also clarifies operational limits, including how this differs from built-in memory, what does not sync automatically, the possibility of sending data to cloud models, and the resource costs of local models.

Similar services

Unknown region · Developer toolsrubi

Knowledge graph and context manager for AI coding agents: Claude Code, Codex, Gemini CLI, Cursor, Copilot and local models.

Unknown region · Developer toolsagentmemoryleaderboard

A public benchmark that compares agent memory systems using the same procedure

Unknown region · MarketingSEOAgent

A local-first SEO tool connecting coding-agent content creation with real search data

Unknown region · Developer toolsctx

Instant recall for coding agents.