# Letta Code

Stateful agents that are like people, with memory, identity, and the ability to learn and adapt

Repository: https://github.com/letta-ai/letta-code
Canonical: https://ross.abutalabs.com/products/letta-code
Homepage: https://www.letta.com/agent
Language: TypeScript
License: Apache-2.0
License Family: permissive
Topics: agent-memory, ai, claude, codex, continual-learning, letta, memgpt, stateful-agents
Last push: 2026-08-26T22:27:20+00:00
Link (homepage): https://www.letta.com/agent
Link (site_page): https://www.letta.com/blog/sleep-time-compute
Link (site_page): https://www.letta.com/about-us

## Health v2 (maintenance only)
Score: 79/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 87, longevity 22
- inputs: {"age_days": 312, "days_push": 7, "days_rel": 7, "gap_med": 0, "n_releases_24m": 290}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3125, forks 372 (observed 2026-08-28T04:07:44.385558+00:00)

## What it is
Letta (formerly MemGPT) is an open-source platform for building stateful AI agents with persistent memory that learn and self-improve over time. It ships as an npm-installable agent harness with an interactive terminal UI, an app server for self-hosting, desktop/web apps, chat channels (Slack, Discord, Telegram), and a TypeScript Agent SDK.

## Use cases
- build stateful AI agents with long-term memory
- create a personal AI assistant that remembers me across sessions
- self-host an agent server with persistent identity and memory
- run an AI coding agent in the terminal
- connect an agent to Slack, Discord, or Telegram
- build agents that learn and improve from experience over time
- schedule proactive background agent tasks

## When to choose
- you need agents with persistent memory and continual learning rather than stateless LLM calls
- you want a self-hosted agent runtime plus ready-made terminal, desktop, and web clients
- you want agents reachable through existing chat channels like Slack or Discord
- you're building TypeScript apps and want an SDK for embedding stateful agents

## When to avoid
- you only need simple stateless LLM API calls without memory
- you need a lightweight library with minimal runtime footprint
- you require the retired Letta V1 API server, which is unsupported and archived
- you want a framework locked to a single model provider's managed stack

## Facets
- artifact type: framework
- maturity: active
- function: agent-framework, llm-inference, chatbot, cli, http-server, mcp
- domain: artificial-intelligence, large-language-models, developer-tools, self-hosted
- platform: cross-platform, windows, cli, self-hosted
- tags: stateful-agents, agent-memory, memgpt, continual-learning, sleep-time-compute, terminal-ui, agent-sdk, chat-channels, ai-agents, nodejs, macos, linux, web-server

## Member repositories
- letta-ai/letta-code (main) score 79
- letta-ai/letta (mirror) score 85
- letta-ai/claude-subconscious (examples) score 75

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:44.385558+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:56:13.384689+00:00, confidence not recorded.
  - readme: https://github.com/letta-ai/letta-code (fetched 2026-08-28T04:07:44.385558+00:00, sha 00c3a2a06472)
  - homepage: https://www.letta.com/agent (fetched 2026-08-29T07:54:28.576585+00:00, sha 219f7b5e10b0)
  - site_page: https://www.letta.com/about-us (fetched 2026-08-29T07:54:28.589181+00:00, sha 321781b099b4)
  - registry_npm: https://registry.npmjs.org/letta-code (fetched 2026-08-29T07:54:28.606801+00:00, sha bd223b9545e7)
  - site_page: https://www.letta.com/blog/sleep-time-compute (fetched 2026-08-29T07:54:28.586543+00:00, sha f0d0cc22f668)
- Data as of 2026-08-30T08:39:29.467469+00:00.