Engram MCP Server
io.github.debragail/engram
Persistent memory for AI agents — store complete conversations and search by meaning.
What is the Engram MCP server?
Engram is an MCP-native memory server that stores complete, uncompressed conversation transcripts and makes them searchable via semantic search. Connect any MCP-compatible client—Claude Desktop, Claude Code, Cursor, Windsurf, Zed—and your agent remembers everything across sessions using embeddings-based semantic search.
Engram provides persistent memory infrastructure for AI agents by storing verbatim conversation transcripts and enabling semantic search across them. It runs entirely on Cloudflare's developer platform and handles multi-tenant isolation, team collaboration, and API key management. Use it to give your AI agent long-term memory that accumulates context across conversations without summarization or compression.
How to install Engram
Copy-paste configuration for popular MCP clients.
AuthorizationrequiredsecretBearer token — use your Engram API key (engram_sk_live_*) or OAuth access token
Tools & capabilities
Tools this server exposes to the agent.
create_conversation— Start a new conversation with optional title, tags, metadataappend_messages— Add messages to an existing conversationsearch— Semantic search across all conversationsget_conversation— Retrieve a conversation with its messageslist_conversations— List conversations with filtering and paginationdelete_conversation— Remove a conversation and its data
Use cases
- Give Claude or Cursor long-term memory of past conversations and decisions across sessions
- Search through accumulated conversation history by semantic meaning to find relevant context
- Build AI agents that maintain persistent context and learn from previous interactions
- Organize conversations with tags and metadata for team collaboration and knowledge management
- Retrieve specific conversations or browse conversation history with filtering and pagination
Engram MCP server FAQ
Engram is a memory server for AI agents that stores complete conversation transcripts verbatim and makes them searchable by meaning using semantic embeddings. It works with any MCP-compatible client like Claude, Cursor, or Windsurf.
Engram offers a free tier with 1,000 messages/month. Paid plans include Pro ($9/mo, 100k messages), Team ($27/seat/mo, 500k messages), and Enterprise (custom pricing).
Use `/plugin marketplace add get-engram/engram` followed by `/plugin install engram@engram`. Restart Claude Code and approve access in the browser—a free account is created automatically via OAuth.
Point your MCP client to the remote server URL `https://mcp.getengram.app/mcp` in your config. For stdio-only clients, use mcp-remote as a bridge. OAuth handles authentication automatically—no API key needed.
Yes, Engram uses OAuth (RFC 9728/8414/7591 with PKCE) for authentication. No API keys to copy—the client discovers the server, registers itself, and signs you in automatically.
Engram uses bge-base-en-v1.5 embeddings generated by Cloudflare Workers AI to index conversations. You can search across all stored conversations by meaning, not just keywords.
README (reference)
Source of truth, from the repository.
Engram
Memory infrastructure for AI agents. Store every conversation verbatim. Search by meaning.
Engram is an MCP-native memory server that stores complete, uncompressed conversation transcripts and makes them searchable via semantic search. Connect any MCP-compatible client — Claude Desktop, Claude Code, Cursor, Windsurf, Zed — and your agent remembers everything across sessions.
Quick start
Claude Code
/plugin marketplace add get-engram/engram
/plugin install engram@engram
Restart Claude Code and approve access in the browser. There is no API key to copy and no config file to edit — the server implements the MCP authorization flow (RFC 9728 / 8414 / 7591, PKCE), so the client discovers it, registers itself, and signs you in. A free account is created as part of signing in.
The plugin also ships a skill that tells Claude when to search memory and what is worth saving, so context accumulates without being asked.
Any other MCP client
Point it at the remote server and let OAuth handle access:
{
"mcpServers": {
"engram": {
"type": "http",
"url": "https://mcp.getengram.app/mcp"
}
}
}
For a client that only speaks stdio, bridge to the remote server with mcp-remote — it drives the same browser OAuth flow, so there is still no key to copy:
{
"mcpServers": {
"engram": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.getengram.app/mcp"]
}
}
}
(An older version of this README suggested npx @getengram/cli mcp; the CLI
has no such command and that configuration never worked.)
How it works
- Verbatim storage — every message stored exactly as sent, no summarization or compression
- Semantic search — find relevant context by meaning using bge-base-en-v1.5 embeddings
- MCP-native — speaks the Model Context Protocol natively, works with any compatible client
- Multi-tenant — per-organization isolation, team seats, and API key management
Architecture
Runs entirely on Cloudflare's developer platform:
- Workers — Hono.js API and MCP server
- D1 — SQLite at the edge for messages and metadata
- Vectorize — semantic search index
- Workers AI — embedding generation
Read the full architecture deep-dive.
MCP tools
Engram exposes 6 tools via MCP:
| Tool | Description |
|---|---|
create_conversation | Start a new conversation with optional title, tags, metadata |
append_messages | Add messages to an existing conversation |
search | Semantic search across all conversations |
get_conversation | Retrieve a conversation with its messages |
list_conversations | List conversations with filtering and pagination |
delete_conversation | Remove a conversation and its data |
See the API reference for full parameters and examples.
Packages
| Package | Description |
|---|---|
apps/mcp-server | Cloudflare Worker — MCP server and REST API |
apps/cli | CLI and MCP bridge (@getengram/cli) |
packages/sdk | TypeScript SDK (@getengram/sdk) |
packages/db | Database queries and migrations |
packages/shared | Shared constants, types, and utilities |
Integration guides
Pricing
| Plan | Price | Messages/month |
|---|---|---|
| Free | $0 | 1,000 |
| Pro | $9/mo | 100,000 |
| Team | $27/seat/mo | 500,000 |
| Enterprise | Custom | Unlimited |
Links
- Website: getengram.app
- Documentation: getengram.app/docs
- Blog: getengram.app/blog
- npm: @getengram/cli
License
Business Source License 1.1 — see LICENSE for details.
Related MCP servers
io.github.debridge-finance/debridge-mcp
Cross-chain and same-chain swaps, fee estimation, and trade execution via AI agents.
io.github.debu-sinha/excalidraw
Security-hardened Excalidraw MCP server with auth, rate limiting, and 14 tools
Stable eight-tool MCP server for changed-UI assurance, authority-aware UI task context, and repair.
Give your AI agents the ability to listen. Microphone capture and speech-to-text.
Decide Policy Notaries
Fail-closed refund, cancellation, return, and trial checks for 100 US subscription vendors.
Google Calendar access for LLM agents: find, create, update, delete events with scheduling intelligence.