# Trove

> Trove is an open-source MCP server for knowledge and memory. Open-source shared memory server where every recalled fact cites its source text. It runs as a hosted remote server, so there is nothing to install locally.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** Web, API, self-hosted
- **Website:** https://mytrove.in
- **Repository:** https://github.com/anunay999/trove
- **MCP Registry name:** io.github.anunay999/trove
- **Categories:** knowledge and memory
- **Listing updated:** 2026-11-14

## About

Trove stores what agents learn in a knowledge graph that several agents can share, so a note saved by Claude Code can be recalled later from Codex or Cursor. Beliefs carry validity intervals: when a fact changes, the old edge is invalidated rather than deleted, which lets you ask what the graph believed at an earlier time.

Evidence is the central idea. Recalled atoms either resolve to cited source text or are explicitly labelled AGENT INFERENCE, and a linter flags nodes with no evidence. The toolset is small: remember, recall, grep, read, connect and forget, with ingest and curation tools for write-scoped keys. recall packs the most relevant memories into a token budget you set. A dashboard shows memory metrics and an explorable graph, and an importer can seed the graph from an Obsidian vault.

You can use the hosted service at mytrove.in with an API key or the Connectors panel, or self-host it, in which case a local instance needs no key when auth is disabled. Optional Claude Code skills teach agents to recall before re-deriving and to supersede rather than delete.

## Features

- **Cited evidence:** Recalled facts trace to source text or are marked as agent inference.
- **Time-aware beliefs:** Old facts are invalidated, not deleted, so past states can be queried.
- **Token-budgeted recall:** recall returns the most relevant memories within a budget you set.
- **Graph dashboard:** Browse memory metrics, lint health and a force-directed graph.
- **Obsidian import:** Seed the graph from a vault with deduplicated re-imports.

## Good for

- Share memory between Claude Code, Codex and Cursor
- Keep decisions and facts across sessions
- Audit which source supports a remembered claim
- Import an Obsidian vault as agent memory
- Self-host a private memory store

## Install

### Claude Code (Terminal)

```sh
claude mcp add --transport http trove https://mytrove.in/mcp
```

### Claude Desktop (Settings → Connectors → Add custom connector)

```sh
https://mytrove.in/mcp
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "trove": {
      "url": "https://mytrove.in/mcp"
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "trove": {
      "type": "http",
      "url": "https://mytrove.in/mcp"
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "trove": {
      "serverUrl": "https://mytrove.in/mcp"
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "trove": {
      "type": "http",
      "url": "https://mytrove.in/mcp"
    }
  }
}
```

## Tools

- `remember`: Save a fact or note.
- `recall`: Retrieve relevant memories within a token budget.
- `grep`: Search memory text.
- `read`: Read a stored item.
- `connect`: Link related items.
- `forget`: Remove an item.

## FAQ

### Can I self-host it?

Yes. It is open source and can be run locally, where no key is needed when auth is off.

### How do I connect Claude Code?

Run claude mcp add with the hosted /mcp URL and a Bearer key created on the dashboard.

### What happens when a fact changes?

The old edge is invalidated rather than deleted, so history is kept.

### Which tools does it expose?

remember, recall, grep, read, connect and forget, plus ingest and curation tools for write-scoped keys.

### Is Trove free?

Yes. Trove is open source, so you can use and self-host it for free.

### How do I add Trove to Claude Code?

Run this command in your terminal: claude mcp add --transport http trove https://mytrove.in/mcp

### What are the best alternatives to Trove?

The closest alternatives to Trove are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

## Alternatives

- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Nautilus Compass Memory](https://ai-powered.com/mcp/nautilus-compass-memory): Nautilus Compass Memory is an open-source MCP server for knowledge and memory.
- [note1](https://ai-powered.com/mcp/note1): note1 is an MCP server for productivity.
- [MedXpert 图书馆连接器（通用版）](https://ai-powered.com/mcp/medxpert): MedXpert 图书馆连接器（通用版） is an open-source MCP server for knowledge and memory.
- [Context Nest](https://ai-powered.com/mcp/context-nest): Context Nest is a freemium MCP server for knowledge and memory.
- [Tulsa Ecosystem Map](https://ai-powered.com/mcp/tulsa-ecosystem-map): Tulsa Ecosystem Map is a free MCP server for knowledge and memory.

---
Source: https://ai-powered.com/mcp/trove-5
