letta 是一个「general」类 AI agent,解决:让 AI 代理在长时间任务中保持上下文记忆并持续优化行为策略。定价:产品页未标明。截至 2026-10-14,KanonAgent 记录到 25.1k 票。本站首次收录于 2026-07-24。
Letta 是一个支持状态记忆的 AI 代理平台，能让 AI 代理长期学习、自我进化，具备持续记忆和任务复盘能力，适用于复杂长期任务。
25.1k 票
Kanon 于 2026 年 7 月 24 日 收录
状态依据:daily change -27192.0 is -3.1 standard deviations from this entity's own prior mean (3628.1±9972.7, baseline excludes this point) — unusual, not necessarily adoption
📈 时间线什么时候变了什么
- 异常 daily change -27192.0 is -3.1 standard deviations from this entity's own prior mean (3628.1±9972.7, baseline excludes this point) — unusual, not necessarily adoption
- 降温 npm_downloads_7d momentum fell to -38389 per 3d from 42326 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to -11396 per 3d from 22607 (<= 40% of the prior window)
- 异常 daily change -27192.0 is -3.1 standard deviations from this entity's own prior mean (3628.1±9972.7, baseline excludes this point) — unusual, not necessarily adoption
- 降温 npm_downloads_7d momentum fell to -37134 per 3d from 42326 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to -10795 per 3d from 22607 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to 0 per 3d from 40656 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to 0 per 3d from 28755 (<= 40% of the prior window)
- 加速 npm_downloads_7d gained 42326 in the last 3d vs 8252 in the 3d before (ratio 5.1x, threshold 1.5x)
- 加速 npm_downloads_7d gained 22607 in the last 3d vs 10046 in the 3d before (ratio 2.3x, threshold 1.5x)
- 降温 npm_downloads_7d momentum fell to 0 per 3d from 40656 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to 0 per 3d from 28755 (<= 40% of the prior window)
- 加速 npm_downloads_7d gained 42326 in the last 3d vs 22358 in the 3d before (ratio 1.9x, threshold 1.5x)
- 加速 npm_downloads_7d gained 22607 in the last 3d vs 10979 in the 3d before (ratio 2.1x, threshold 1.5x)
- 加速 npm_downloads_7d gained 28755 in the last 3d vs 14281 in the 3d before (ratio 2.0x, threshold 1.5x)
- 加速 npm_downloads_7d gained 40656 in the last 3d vs 24028 in the 3d before (ratio 1.7x, threshold 1.5x)
- 降温 npm_downloads_7d momentum fell to 8252 per 3d from 32867 (<= 40% of the prior window)
- 加速 npm_downloads_7d gained 25684 in the last 3d vs 14281 in the 3d before (ratio 1.8x, threshold 1.5x)
- 降温 npm_downloads_7d momentum fell to 8252 per 3d from 32867 (<= 40% of the prior window)
- 加速 npm_downloads_7d gained 33896 in the last 3d vs 18591 in the 3d before (ratio 1.8x, threshold 1.5x)
- 降温 npm_downloads_7d momentum fell to 4831 per 3d from 18087 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to 32867 per 3d from 109573 (<= 40% of the prior window)
- 降温 npm_downloads_7d momentum fell to 19020 per 3d from 84863 (<= 40% of the prior window)
- 加速 npm_downloads_7d gained 33896 in the last 3d vs 18591 in the 3d before (ratio 1.8x, threshold 1.5x)
🔎 已知 / 未知
23 个字段里 15 个未知
| 可机器调用 | — | 未知 | — |
| 开源 | 1 | 已核实 | 1 条引文 |
| 可自托管 | — | 未知 | — |
| 需自带 key | — | 未知 | — |
| 自主度 | — | 未知 | — |
| 计费模式 | — | 未知 | — |
| 集成面 | slack | 推断 | 1 条引文 |
「未知」= 我们没核实过,**不等于「否」**。硬性筛选永远不会把未知当成否。
🤖 Agent 拆解 · general
解决什么场景让 AI 代理在长时间任务中保持上下文记忆并持续优化行为策略
目标市场AI 工程团队、开发者
前置条件开源
集成面slack
牵引 · 为什么在涨Product Hunt 票数 23,942，GitHub 星标增长快
为什么值得关注
突破传统 AI 会话无记忆的限制，实现类人长期记忆与自我改进，是构建真正智能代理的关键。
证据引文判定所依据的原文 · 逐字引用
“public GitHub repository (README fetched)”— structural
“Platform for stateful agents: AI with advanced memory that can learn and self-improve over time.”— description
“via channels like Slack”— readme
信号来源: GitHub
访问官网 →
📛 官方徽章挂到官网 / README
你是 letta 的作者?上面选样式,嵌入代码实时更新;要贴合站点配色可在 URL 上加 bg= / fg= / accent=(十六进制色)。徽章链回本页。
手机端点「分享」直达微信/朋友圈/小红书;桌面端用「复制文案」后到 App 内粘贴发布
常见问题
letta 是什么?
Letta 是一个支持状态记忆的 AI 代理平台，能让 AI 代理长期学习、自我进化，具备持续记忆和任务复盘能力，适用于复杂长期任务。
letta 解决什么问题?
让 AI 代理在长时间任务中保持上下文记忆并持续优化行为策略
letta 为什么值得关注?
突破传统 AI 会话无记忆的限制，实现类人长期记忆与自我改进，是构建真正智能代理的关键。
letta 是开源的吗?
是。letta 为开源项目。
letta 能接哪些工具?
slack
letta 有多少人在用?
KanonAgent 记录到:25.1k 票(本站首次收录于 2026-07-24)。
letta 有什么替代品?
KanonAgent 库内的同类 agent:Stealth Company、Anonymous Startup、hermes-agent、AutoGPT、Anonymous Startup、Confidential Startup。