letta-ai/letta-code
The memory-first coding agent
TypeScript · ★ 1,535 · Fork 158 · 在 GitHub 查看
在手机上浏览源码、订阅每日 AI 趋势日报与项目架构深度解读:获取 CopoHub 客户端(iOS & HarmonyOS)→
GitHub Trending 上榜记录
letta-ai/letta-code 已在 GitHub Trending 每日榜单上榜 4 天,最高排名第 7,首次上榜 2026-02-14,最近一次上榜 2026-02-17。
| 日期 | 排名 | Star | 当日增长 |
|---|---|---|---|
| 2026-02-17 | #7 | 1,535 | +22 |
| 2026-02-16 | #7 | 1,432 | +51 |
| 2026-02-15 | #9 | 1,300 | +55 |
| 2026-02-14 | #9 | 1,186 | +30 |
同期上榜的项目
- p-e-w/heretic – Training-free language model modification: Remove censorship, suppress slop, reduce sycophancy, and more
- nautechsystems/nautilus_trader – Production-grade Rust-native trading engine with deterministic event-driven architecture
- seerr-team/seerr – Open-source media request and discovery manager for Jellyfin, Plex, and Emby.
- obra/superpowers – An agentic skills framework & software development methodology that works.
- rowboatlabs/rowboat – Open-source AI coworker, with memory
- steipete/gogcli – Google Suite CLI: Gmail, GCal, GDrive, GContacts.
letta-code 项目深度剖析与商业评估报告
letta-code 项目深度剖析与商业评估报告
报告日期:2026-08-22 | 分析对象:letta-ai/letta-code v(TypeScript / Apache 2.0) 元数据快照:3.1k Stars / 363 Forks / 238 Open Issues / 最近推送:2026-08-22
---
1. 核心定位与商业落地价值
1.1 根本痛点:LLM 的"金鱼记忆"与工具化困境
当前 LLM 应用面临两大结构性矛盾:上下文窗口的物理上限与长期任务的持续学习需求。letta-code 的切入点并非又一个 Agent 框架,而是直接挑战"Agent 即无状态函数调用链"这一主流范式,提出有状态、可自我演进的 Agent 运行时(Stateful Agent Harness)。
其核心主张是:Agent 不应是工具,而应是"具有身份连续性的实体"。这一定位直接回应了企业级 AI 应用中知识沉淀断裂的问题——传统 RAG 方案只能解决"检索外部知识",无法解决"Agent 自身经验跨会话累积"。
1.2 商业落地场景矩阵
在 README 中展示上榜徽章
[](https://github.fq6825.top/repo/letta-ai/letta-code)