• Which tools were called
• Where errors occurred
A clearer view of how your agents work together, and where to start debugging when they don’t.
EverMind-AI/raven
Has anyone tried Raven from Evermind? Looking into it and it seems alot like deepseek superagent before DSH superagent..
Kinda shocked I can't find anything on reddit about this. Really intrigued by it, but want to find some anecdotal stuff from the community before I go through the labor of tinkering and testing. Anyone tried this thing? EverMind-AI/Raven
1 point · 2 comments
Raven is an open source agent host from EverMind AI. It plans a task graph and hands each part to a specialist agent, built in or third party. Free and open source under Apache 2.0. EverMind-AI/Raven t.co
23 views
Beyond Gain: Measuring Skill Internalization for Agent Skill Retirement
Yanze Wang, Pengfei Yao, Tianyi Sun, Hongda Chen, Yunyun Han, Jun Sun, Chuanrui Hu, Yafeng Deng
arxiv.org [𝚌𝚜.𝙰𝙸]
💬Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… t.co
85 views
SkillRep: Characterizing and Understanding the Representation Design Space of Agent Skill Retrieval
Tianyi Sun, Pengfei Yao, Yanze Wang, Hongda Chen, Yunyun Han, Jun Sun, Chuanrui Hu, Yafeng Deng
arxiv.org [𝚌𝚜.𝙸𝚁]
💬Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… t.co
1 like · 42 views
Beyond Gain: Measuring Skill Internalization for Agent Skill Retirement
LLM agents retrieve skills from public hubs that hold up to about 100,000 entries and serve many models. As models improve, a hub must find the skills a given model has internalized, meaning it can do without them, and retire them for that model, yet internalization has not been measured with known reliability. Published rules that compare runs with and without a skill retire on the gain alone. The gain is near zero both for an internalized skill and when neither the model nor the skill delivers the capability, a state we call uncovered. We instead withhold the skill, score the model on a task synthesized from it, and trust a high unaided score only when another model gains from the skill, since nearly half of high scores come from tasks no model benefits from. Under four models, a published gain rule applied with one task per skill removes over half of the uncovered skills. On SkillsBench, 28 of its 31 retirements rest on probes no run passed in full, while retiring on the unaided score costs no detectable reward. Unaided verdicts reproduce across tasks more often than the gain rule's (Cohen's kappa 0.42 against 0.23) but differ across models, so a hub must read them per model. We release 1,251 skills with tasks and item scores under four models.
Comments: 12 pages. Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… Data: huggingface.co
SkillRep: Characterizing and Understanding the Representation Design Space of Agent Skill Retrieval
Retrieving the right skill from a large library is essential for LLM agents, yet requests and skills are written for different purposes and in different words. Existing retrievers narrow this request-skill gap with one or two representation choices each, so it is unknown whether their gains hold in combination. No prior work measures the gap itself, so it is also unclear which part of it each choice narrows. We define a representation design space by five choices -- encoder fine-tuning, query-side and document-side expansion, background projection and a lexical channel -- and evaluate them together for the first time, as modules of one retriever, SkillRep. With a 0.6B encoder and LLM expansion, SkillRep improves MRR@10 on SRA-Bench over the strongest same-size retriever by 7.02 points with the official skill pool and 8.79 with UniPool, a 111,247-skill pool we build from public registries, and is best on SkillsBench. The value of each choice depends on the others. Fine-tuning, for instance, reverses the ranking of the four training-free modules and turns them from complementary into overlapping. To understand why, we measure the gap with three quantities: background, the embedding component that does not separate skills; unsaid vocabulary, the gold skill's content a request leaves unsaid; and dilution, the information lost when a skill is encoded into a single vector. They show which part each module narrows and by what mechanism. This reveals that modules narrowing the same part compete for it, while those narrowing different parts complement each other.
Comments: 12 pages, 3 figures. Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil…
Raven: One Host Agent for Many AI Agents
Raven is an open source agent host from EverMind AI. It plans a task graph and hands each part to a specialist agent, built in or third party. Free and open source under Apache 2.0. EverMind-AI/Raven
180 views
EverOS 本身是 GitHub 上 1.3w 星的开源 Agent 记忆系统，我这次用的 EverOS Cloud 是官方托管的云端版，不用自己搭服务，直接一行 MCP 命令就能挂进 Claude Code、Codex 或 Cursor 里用。
EverOS Cloud 入口 👉 everos.evermind.ai
GitHub 开源地址 👉 EverMind-AI/EverOS
个人日常用免费额度其实就够跑了；如果项目多或者团队一起用，他们最近正好有活动：现在订阅 Pro，每月额度变 3 倍，连享 6 个月。
4 likes · 1 repost · 172 views
💾 盛大团队孵化的开源项目 EverOS，给 Claude Code、Codex 这类 AI agent 装上长期记忆，踩过的坑下次不用再踩。
背后是陈天桥创办的盛大投资集团，EverOS 由其研究团队孵化。GitHub 已经 1.3 万 star，10 月中旬刚更新到 1.4.2。
GitHub：EverMind-AI/EverOS
用 Claude Code 久了都遇到过：新开一个会话，它把上周刚踩过的坑原样再踩一遍。你只好把教训一条条手写进 CLAUDE.md，文件越写越长，最后自己都不敢删。
EverOS 把这件事交给记忆层来做。你的偏好和经历记在一边，agent 每次干活的过程记成一条 Case。同类 Case 攒多了，它会在两次会话之间把反复成功的做法提炼成 Skill，下次遇到类似任务先查一遍再动手。
记忆本身全是 Markdown 文件，能直接打开改，也能用 git 看每次变了什么。本地跑只要 SQLite 加 LanceDB，不用另装 MongoDB、Redis 那一套。
Claude Code 有现成插件，Codex 和 Cursor 走 MCP 接入。云端和自托管是同一套 API，先用云版试，想搬回自己机器也不用改代码。官方用例里已经有人拿它做了 Rokid 眼镜上的助手，还有给阿尔茨海默症患者用的记忆助手。
agent 不缺聪明，缺的是记性。
1 like · 529 views
Mentions
• Which tools were called • Where errors occurred A clearer view of how your agents work together, and where to start debugging when they don’t. EverMind-AI/raven
Has anyone tried Raven from Evermind? Looking into it and it seems alot like deepseek superagent before DSH superagent.. Kinda shocked I can't find anything on reddit about this. Really intrigued by it, but want to find some anecdotal stuff from the community before I go through the labor of tinkering and testing. Anyone tried this thing? EverMind-AI/Raven
Raven is an open source agent host from EverMind AI. It plans a task graph and hands each part to a specialist agent, built in or third party. Free and open source under Apache 2.0. EverMind-AI/Raven t.co
Beyond Gain: Measuring Skill Internalization for Agent Skill Retirement Yanze Wang, Pengfei Yao, Tianyi Sun, Hongda Chen, Yunyun Han, Jun Sun, Chuanrui Hu, Yafeng Deng arxiv.org [𝚌𝚜.𝙰𝙸] 💬Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… t.co
SkillRep: Characterizing and Understanding the Representation Design Space of Agent Skill Retrieval Tianyi Sun, Pengfei Yao, Yanze Wang, Hongda Chen, Yunyun Han, Jun Sun, Chuanrui Hu, Yafeng Deng arxiv.org [𝚌𝚜.𝙸𝚁] 💬Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… t.co
Beyond Gain: Measuring Skill Internalization for Agent Skill Retirement LLM agents retrieve skills from public hubs that hold up to about 100,000 entries and serve many models. As models improve, a hub must find the skills a given model has internalized, meaning it can do without them, and retire them for that model, yet internalization has not been measured with known reliability. Published rules that compare runs with and without a skill retire on the gain alone. The gain is near zero both for an internalized skill and when neither the model nor the skill delivers the capability, a state we call uncovered. We instead withhold the skill, score the model on a task synthesized from it, and trust a high unaided score only when another model gains from the skill, since nearly half of high scores come from tasks no model benefits from. Under four models, a published gain rule applied with one task per skill removes over half of the uncovered skills. On SkillsBench, 28 of its 31 retirements rest on probes no run passed in full, while retiring on the unaided score costs no detectable reward. Unaided verdicts reproduce across tasks more often than the gain rule's (Cohen's kappa 0.42 against 0.23) but differ across models, so a hub must read them per model. We release 1,251 skills with tasks and item scores under four models. Comments: 12 pages. Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil… Data: huggingface.co
SkillRep: Characterizing and Understanding the Representation Design Space of Agent Skill Retrieval Retrieving the right skill from a large library is essential for LLM agents, yet requests and skills are written for different purposes and in different words. Existing retrievers narrow this request-skill gap with one or two representation choices each, so it is unknown whether their gains hold in combination. No prior work measures the gap itself, so it is also unclear which part of it each choice narrows. We define a representation design space by five choices -- encoder fine-tuning, query-side and document-side expansion, background projection and a lexical channel -- and evaluate them together for the first time, as modules of one retriever, SkillRep. With a 0.6B encoder and LLM expansion, SkillRep improves MRR@10 on SRA-Bench over the strongest same-size retriever by 7.02 points with the official skill pool and 8.79 with UniPool, a 111,247-skill pool we build from public registries, and is best on SkillsBench. The value of each choice depends on the others. Fine-tuning, for instance, reverses the ranking of the four training-free modules and turns them from complementary into overlapping. To understand why, we measure the gap with three quantities: background, the embedding component that does not separate skills; unsaid vocabulary, the gold skill's content a request leaves unsaid; and dilution, the information lost when a skill is encoded into a single vector. They show which part each module narrows and by what mechanism. This reveals that modules narrowing the same part compete for it, while those narrowing different parts complement each other. Comments: 12 pages, 3 figures. Code: github.com/EverMind-AI/SkillCorpus/tree/main/skil…
Raven: One Host Agent for Many AI Agents Raven is an open source agent host from EverMind AI. It plans a task graph and hands each part to a specialist agent, built in or third party. Free and open source under Apache 2.0. EverMind-AI/Raven
EverOS 本身是 GitHub 上 1.3w 星的开源 Agent 记忆系统，我这次用的 EverOS Cloud 是官方托管的云端版，不用自己搭服务，直接一行 MCP 命令就能挂进 Claude Code、Codex 或 Cursor 里用。 EverOS Cloud 入口 👉 everos.evermind.ai GitHub 开源地址 👉 EverMind-AI/EverOS 个人日常用免费额度其实就够跑了；如果项目多或者团队一起用，他们最近正好有活动：现在订阅 Pro，每月额度变 3 倍，连享 6 个月。
💾 盛大团队孵化的开源项目 EverOS，给 Claude Code、Codex 这类 AI agent 装上长期记忆，踩过的坑下次不用再踩。 背后是陈天桥创办的盛大投资集团，EverOS 由其研究团队孵化。GitHub 已经 1.3 万 star，10 月中旬刚更新到 1.4.2。 GitHub：EverMind-AI/EverOS 用 Claude Code 久了都遇到过：新开一个会话，它把上周刚踩过的坑原样再踩一遍。你只好把教训一条条手写进 CLAUDE.md，文件越写越长，最后自己都不敢删。 EverOS 把这件事交给记忆层来做。你的偏好和经历记在一边，agent 每次干活的过程记成一条 Case。同类 Case 攒多了，它会在两次会话之间把反复成功的做法提炼成 Skill，下次遇到类似任务先查一遍再动手。 记忆本身全是 Markdown 文件，能直接打开改，也能用 git 看每次变了什么。本地跑只要 SQLite 加 LanceDB，不用另装 MongoDB、Redis 那一套。 Claude Code 有现成插件，Codex 和 Cursor 走 MCP 接入。云端和自托管是同一套 API，先用云版试，想搬回自己机器也不用改代码。官方用例里已经有人拿它做了 Rokid 眼镜上的助手，还有给阿尔茨海默症患者用的记忆助手。 agent 不缺聪明，缺的是记性。