Letta MCP Server

Official Letta MCP server enabling AI agents to interact with Letta (formerly MemGPT) — the stateful agent framework with persistent memory management — creating and managing Letta agents, accessing long-term memory stores, adding and searching memories, running stateful agent conversations, and integrating Letta's memory-augmented agent infrastructure into MCP-based workflows.

Evaluated Mar 06, 2026 (0d ago) vcurrent
Homepage ↗ Repo ↗ AI & Machine Learning letta memgpt memory mcp-server agent-memory long-term-memory stateful-agents
⚙ Agent Friendliness
75
/ 100
Can an agent use this?
🔒 Security
81
/ 100
Is it safe for agents?
⚡ Reliability
70
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
72
Documentation
75
Error Messages
72
Auth Simplicity
80
Rate Limits
78

🔒 Security

TLS Enforcement
90
Auth Strength
80
Scope Granularity
75
Dep. Hygiene
78
Secret Handling
82

HTTPS for Cloud. API key auth. Memory data may be sensitive — review retention. Official Letta MCP.

⚡ Reliability

Uptime/SLA
75
Version Stability
68
Breaking Changes
68
Error Recovery
70
AF Security Reliability

Best When

An agent needs to interact with Letta's stateful memory infrastructure — for building memory-augmented assistants, managing persistent agent state, or integrating MemGPT-style memory into larger agent systems.

Avoid When

You need simple stateless LLM calls or don't have a Letta server running — the overhead isn't worthwhile for short-lived agent tasks.

Use Cases

  • Creating persistent stateful agents with long-term memory from agent orchestration systems
  • Querying and updating agent memory stores from memory management agents
  • Building memory-augmented AI assistants from personal assistant agents
  • Running Letta agents as sub-agents in larger MCP-based systems from orchestration agents
  • Integrating MemGPT-style memory management into existing agent workflows from migration agents
  • Managing multiple stateful agent personas from agent management agents

Not For

  • Simple stateless LLM calls (use direct LLM MCPs for stateless queries)
  • Teams without Letta server deployed (requires local or cloud Letta instance)
  • Short-lived agents where memory persistence adds unnecessary overhead

Interface

REST API
Yes
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
Yes
Webhooks
No

Authentication

Methods: api_key
OAuth: No Scopes: No

Letta API key required for Letta Cloud; local Letta server may not require auth. LETTA_BASE_URL configures endpoint. Self-hosted Letta is also supported.

Pricing

Model: freemium
Free tier: Yes
Requires CC: No

Letta is open source — self-hosted is completely free. Letta Cloud offers managed hosting with a free tier. MCP server is free.

Agent Metadata

Pagination
cursor
Idempotent
Partial
Retry Guidance
Not documented

Known Gotchas

  • Letta server (local or Cloud) must be running and accessible before MCP can connect
  • Letta renamed from MemGPT — some docs and examples still use old MemGPT naming
  • Agent and memory management can be complex — Letta's stateful architecture has learning curve
  • Memory search quality depends on embedding model configured in Letta
  • Official Letta MCP but Letta is a fast-moving project — API may evolve significantly
  • Self-hosted Letta requires LLM provider API keys (OpenAI, Anthropic, etc.) configured separately

Alternatives

Full Evaluation Report

Detailed scoring breakdown, competitive positioning, security analysis, and improvement recommendations for Letta MCP Server.

$99

Scores are editorial opinions as of 2026-03-06.

5190
Packages Evaluated
26151
Need Evaluation
173
Need Re-evaluation
Community Powered