Tier 5 · MCP & tool frameworks MCP & tooling · fetched 2026-07-02

Letta MCP Agent Framework

An MCP server integrating Letta's agent memory capabilities with VS Code Agent Mode - self-improving agents whose memory lives behind MCP tools.

View source on GitHub

Key takeaways

01
Memory exposed as MCP tools makes it portable across hosts

From the archive

Verbatim excerpts mined from our local archive of this repository — the prompts, schemas, and patterns worth stealing.

Prompt

Reusable memory storage and recall prompt blocks

/Letta.remember input:"User prefers detailed technical explanations with code examples"

/Letta.remember input:"Implemented OAuth integration using GitHub provider" tags:["authentication", "github", "oauth", "security"]

/Letta.recall query:"What are the user's coding preferences?"

/Letta.recall query:"How did we solve the authentication issue last time?"

/Letta.recall query:"React component patterns we've used" limit:3

A prompt library for agent self-memory: store decisions with tags at write time, recall with natural-language questions at read time.

agent_prompts.md

MCP & tools

VS Code MCP server registration with capabilities

{
  "id": "letta-mcp",
  "url": "http://localhost:4000",
  "title": "Letta Agent MCP",
  "description": "Letta-integrated MCP server for agent memory and reflection",
  "auth": { "type": "none" },
  "capabilities": [
    "memory_storage",
    "memory_recall",
    "agent_reflection",
    "background_tasks"
  ],
  "endpoints": ["/remember", "/recall", "/reflect", "/status"]
}

A minimal MCP server manifest exposing memory as three verbs - remember, recall, reflect - the reflect endpoint runs background self-analysis.

.vscode/mcp.json