Filter
Exclude
Time range
-
Near
lucas henrique retweeted
Your agents don't talk to each other. Claude Code ships it. Cursor reinvents it. Codex starts from zero. Hivemind is the layer that lets them learn from each other.
4
44
66
760,472
Aaron Hughes 🇮🇪 retweeted
Continual Learning for Claude Code, Cursor and Codex with Hivemind: • Every run gets traced (tools, files, outcomes) • Successful traces get promoted to skills • Next session loads them automatically Now Open Source for Claude Code, Codex, Cursor, Hermes, Pi.
39
26
437,660
🛠️ 开源工具推荐:《Hivemind》—— 给所有 AI Coding Agent 共享一个大脑,从真实轨迹里自动提炼技能 用了好几个 AI Coding Agent 的人大概都遇过这件事:每个工具学到的东西都锁在自己的上下文里,换一个工具就得从头来,例如Claude Code 里用顺手的工作流,在 Codex 或 Cursor 里根本不知道。 Hivemind 解决的是这个问题:它作为所有 Agent 共享的记忆层,自动捕获会话里的 prompt、工具调用和响应轨迹,把重复出现的高质量模式挖掘出来,转化成可复用的 SKILL.md 文件,在团队和工具之间传播。来自 Activeloop(Y Combinator 支持),开源创作者的 Deep Lake 向量数据库同团队。 核心特性: 1. 自动模式挖掘:从你的实际使用轨迹里发现什么值得保存,不需要手动整理 2. 跨 Agent 技能传播:SKILL.md 格式,Claude Code、Codex、Cursor、Hermes Agent 等全支持 3. 混合检索(词法 语义):关键词精确查找和语义模糊匹配双模式覆盖 4. SQL 虚拟文件系统:结构化存储,记忆可查询、可版本化、可迁移 5. 会话摘要 Wiki 自动生成:长对话浓缩成可索引的知识片段 6. 基准验证(LoCoMo):比基线便宜 25%、每个问题少用 1.7 倍 token、减少 31% 交互轮次 项目特别适合同时用多个 AI Coding Agent 工作、希望工具之间能互相「学习」而不是各自孤立运行的开发者和工程团队。目前已获得 929 stars ⭐,是当前跨 Agent 共享记忆与技能传播领域最完整的开源实现之一。 与单个 Agent 的内置记忆系统(如 Claude Code 的 CLAUDE.md)相比,Hivemind 的核心差异在于「跨工具」:不是 Claude Code 自己的记忆,而是所有工具共享的技能库,换工具不等于从头开始,团队的工程智慧可以在工具间传播。 github.com/activeloopai/hive… #AIAgent #ClaudeCode #VibeCoding #codex #AIEngineering
4
9
60
4,228
Coach Glen retweeted
Continual learning and updated skills with SkillOpt. • Every run gets traced (tools, files, outcomes) • Successful traces get promoted to skills • Skills are updated and improved Continual learning now Open Source for Claude Code, Codex, Cursor, Hermes, Pi.
1
16
7
71,054
WRS retweeted
Turn repeated prompts into auto-generated skills. We built Hivemind to turns your repeated prompts into auto-generated skills. Up-level your team's agents.
2
54
323
4,631,005
Very good revamp man.
10
Owen Barford retweeted
Create a knowledge graph of your codebase. • Every run gets traced (tools, files, outcomes) • Successful traces get promoted to skills • Knowledge is shared across your team's agents Continual learning now Open Source for Claude Code, Codex, Cursor, Hermes, Pi.
1
6
20
360,872
2026/6/13 8:24 JST 11955kHz #CRI China Radio Int. Kunming-Anning, China English Program SINPO 35343 昆明市安寧市(Kunming-Anning)CRIの #英語放送 昆明市安寧市(あんねいし)は中国雲南省の省都である昆明市の管轄下にある県級市 RX: RTL-SDR BLOG V4 LZ1AQ ActiveLoop #SWL
2
150
2026/6/13 8:13 JST 11790kHz #CRI China Radio Int, Xianyang 🇨🇳 English program SINPO 45455 咸陽(Xianyang) CRIの #英語放送 咸陽は、中国・陝西省(せんせいしょう)に位置する歴史ある大都市 RX: RTL-SDR BLOG V4 LZ1AQ ActiveLoop #SWL
3
145
2026/6/13 8:03 JST 11690kHz #CRI China Radio Int. Beijing, China 🇨🇳 English program SINPO 55555 北京CRIの #英語放送 RX: RTL-SDR BLOG V4 LZ1AQ ActiveLoop #SWL
2
144
Huge improvement. The new version feels much more engaging.
15
lucas henrique retweeted
How Hivemind turns Hermes runs into a compounding asset: • Every run gets traced (tools, files, outcomes) • Successful traces get promoted to skills • Next session loads them automatically Continual learning now Open Source for Claude Code, Codex, Cursor, Hermes, Pi.
25
45
303
1,085,250