R
rhtevan/agentfs
Shared agent scaffolding with skills, knowledge bundles, and cross-agent context for AI agents (Goose, Hermes, Claude Code, etc.)
★ 2·更新 2026-11-18
在 GitHub 查看 ↗链接已于 2026-11-17 核查 · HTTP 200 · 打开的是厂商自己的官网
- ★ 数取值
- 2 ★ · rhtevan/agentfs · 取自 GitHub 搜索 API · (2026-11-19 20:00:48 核对)
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型:开源项目
- 收录与活跃
- 收录于 2026-11-17 · 最近见到 2026-11-20 · 共见到 4 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 2 ★ · 1 个来源 · repo
常见问题(答案都来自我们的记录)
它适合谁 / 用来做什么?
我们把它归在这些标签下: 开源项目。标签来自条目自身,不是我们强加的分类。
谁维护它、还在更新吗?
仓库
rhtevan/agentfs;我们收录于 2026-11-17;最近见到 2026-11-20。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。README 原文(英文,未翻译) · 在 GitHub 读完整 README →
permalink: /
以下摘录自该仓库的 README(片段)
# AgentFS (agent-filesystem)
Shared agent scaffolding with skills, knowledge bundles, and cross-agent context.
Part of the agent-x trilogy:
AgentFS (agent-filesystem) — foundation file structures for agents ← this project
AgentFLOW (agent-flow) — deterministic workflow orchestration
AgentBOX (agent-box) — secure enterprise sandbox for agent sessions
AgentFS is a structured filesystem convention for AI agents (Goose, Hermes, Claude Code, etc.) that enables persistent memory, reusable skills, and shared knowledge across agents and sessions.
## Quick Navigation
| Resource | Path | What's Inside |
| Skills catalog | skills/index.md | All available skills — signal phrases, tags, descriptions |
| Knowledge bundles | knowledge/index.md | Curated knowledge — concepts, patterns, postmortems |
| Deployment profiles | skills/hosted-model-ctl/references/deployment-profiles.md | Self-hosted model profiles — fitting scores, speed, hardware, trade-offs |
| Activity log | log.md | Reverse-chronological change history |
| Design spec | skills/agentfs-setup/references/design-spec.md | Architecture, scope model, layer definitions |
## Scope Definitions
AgentFS operates in three scopes. These definitions are canonical.
| Scope | Root Path | Resolves To | Purpose |
| USER | ~/.agents/ | /home/ /.agents/ | Machine-wide shared library: skills and knowledge visible across all projects and agents |
| PROJECT | ./.agents/ | /.agents/ | Per-repository agent workspace: identity, profiles, memories, and project-scoped skills |
| LITE | ./.agents/ | /.agents/ | Minimal per-repo workspace for small-context models: identity and memories only |
> Rule of thumb: USER scope = ~/.agents/. PROJECT scope = ./.agents/ (CWD). LITE scope = ./.agents/ at a remote path, managed externally.
> This repository is a USER scope AgentFS instance.
## Getting Started
### Step 1: Set Up USER Scope (~/.agents/)
Choose one of two paths:
#### Path A: Full Install (recommended)
Clone this repo directly into ~/.agents/:
bash
git clone https://github.com/rhtevan/agentfs.git ~/.agents
You get the complete skill library, knowledge bundles, and structural
scaffolding — ready to use immediately.
#### Path B: Minimal Install
For a clean, empty ~/.agents/ where you cherry-pick skills:
1. Clone the repo to a staging location:
bash
git clone https://github.com/rhtevan/agentfs.git ~/repos/agentfs
2. Make the staging location visible to your agent (e.g., add
~/repos/agentfs/skills/ to the agent's skill search paths
— see the relevant agent setup skill for details).
3. Ask your agent to run the agentfs-setup skill with USER scope:
> "Set up AgentFS in USER scope"
The agent will load the agentfs-setup skill and scaffold an
empty ~/.agents/ with skills/, knowledge/, index.md,
and log.md.
4. Cherry-pick specific skills using the skill-merge skill or
manual copy.
### Step 2: Configure Your Agent
#### With Goose
Load the goose-agentfs-setup skill to configure Goose for AgentFS:
Context files: Registers CLAUDE.md, .goosehints, etc. in
Goose's CONTEXT_FILE_NAMES
System prompt optimization: Generates a custom system.md that
places AGENTS.md rules at the primacy zone (top of system prompt)
for improved compliance on smaller models. Includes drift detection
Shared agent scaffolding with skills, knowledge bundles, and cross-agent context.
Part of the agent-x trilogy:
AgentFS (agent-filesystem) — foundation file structures for agents ← this project
AgentFLOW (agent-flow) — deterministic workflow orchestration
AgentBOX (agent-box) — secure enterprise sandbox for agent sessions
AgentFS is a structured filesystem convention for AI agents (Goose, Hermes, Claude Code, etc.) that enables persistent memory, reusable skills, and shared knowledge across agents and sessions.
## Quick Navigation
| Resource | Path | What's Inside |
| Skills catalog | skills/index.md | All available skills — signal phrases, tags, descriptions |
| Knowledge bundles | knowledge/index.md | Curated knowledge — concepts, patterns, postmortems |
| Deployment profiles | skills/hosted-model-ctl/references/deployment-profiles.md | Self-hosted model profiles — fitting scores, speed, hardware, trade-offs |
| Activity log | log.md | Reverse-chronological change history |
| Design spec | skills/agentfs-setup/references/design-spec.md | Architecture, scope model, layer definitions |
## Scope Definitions
AgentFS operates in three scopes. These definitions are canonical.
| Scope | Root Path | Resolves To | Purpose |
| USER | ~/.agents/ | /home/ /.agents/ | Machine-wide shared library: skills and knowledge visible across all projects and agents |
| PROJECT | ./.agents/ | /.agents/ | Per-repository agent workspace: identity, profiles, memories, and project-scoped skills |
| LITE | ./.agents/ | /.agents/ | Minimal per-repo workspace for small-context models: identity and memories only |
> Rule of thumb: USER scope = ~/.agents/. PROJECT scope = ./.agents/ (CWD). LITE scope = ./.agents/ at a remote path, managed externally.
> This repository is a USER scope AgentFS instance.
## Getting Started
### Step 1: Set Up USER Scope (~/.agents/)
Choose one of two paths:
#### Path A: Full Install (recommended)
Clone this repo directly into ~/.agents/:
bash
git clone https://github.com/rhtevan/agentfs.git ~/.agents
You get the complete skill library, knowledge bundles, and structural
scaffolding — ready to use immediately.
#### Path B: Minimal Install
For a clean, empty ~/.agents/ where you cherry-pick skills:
1. Clone the repo to a staging location:
bash
git clone https://github.com/rhtevan/agentfs.git ~/repos/agentfs
2. Make the staging location visible to your agent (e.g., add
~/repos/agentfs/skills/ to the agent's skill search paths
— see the relevant agent setup skill for details).
3. Ask your agent to run the agentfs-setup skill with USER scope:
> "Set up AgentFS in USER scope"
The agent will load the agentfs-setup skill and scaffold an
empty ~/.agents/ with skills/, knowledge/, index.md,
and log.md.
4. Cherry-pick specific skills using the skill-merge skill or
manual copy.
### Step 2: Configure Your Agent
#### With Goose
Load the goose-agentfs-setup skill to configure Goose for AgentFS:
Context files: Registers CLAUDE.md, .goosehints, etc. in
Goose's CONTEXT_FILE_NAMES
System prompt optimization: Generates a custom system.md that
places AGENTS.md rules at the primacy zone (top of system prompt)
for improved compliance on smaller models. Includes drift detection
收录徽章(可选)
想把收录标记放进 README 就复制这行,可选,不影响收录。
[](https://anclum.com/zh/s/rhtevan-agentfs/)
- 来源
- github · https://github.com/rhtevan/agentfs
- 按领域浏览
- 开源项目
- Traction
- 2 ★
- 收录于
- 今天
更多条目
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。