官方说明开源项目7 天前
aget-framework/aget
AGET Framework - specification for building self-built Personal AGI Systems with governed CLI agents
在 GitHub 查看 ↗直链 · 无中转页
- ★ 数取值
- 12 ★ · aget-framework/aget · 取自 GitHub 搜索 API · （2026-10-16 20:00:20 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 2026-10-10 · 最近见到 2026-10-10 · 共见到 2 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 12 ★ · 1 个来源 · repo
常见问题（答案都来自我们的记录）
怎么安装？
我们【没有】从该仓库取到安装命令（我们的 install 覆盖率只有约 4%）。请看它的 README 或 releases 页。
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
aget-framework/aget；我们收录于 2026-10-10；最近见到 2026-10-10。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。第三方质量标记
AGET Framework
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
> Persistent domain intelligence for governed agentic work
The Problem
Your CLI agents lose context between sessions. Knowledge resets daily. Agents in the same fleet can't learn from each other. You have no confidence that what worked yesterday still works today.
AGET fixes this. It gives your agents persistent knowledge, shared memory, fleet coordination, and human-supervised governance. Your agents accumulate domain expertise that compounds across sessions.
**New to AGET?** Listen to the audio introduction — a narrated walkthrough of session rhythm, key skills, and fleet creation.
v3.36.0 release candidate
See the release notes and migration handoff. Availability and receiver acceptance require separate verification.
The migration kit's test-body witness requires Python 3.12+ (sys.monitoring); on Python 3.10/3.11 a run needing that witness returns INCONCLUSIVE by design (weekly-train:R15).
Quick Start
Start with the Supervisor template. It coordinates your fleet and can create new agents.
See GETTING_STARTED.md for archetype selection and setup guidance. Follow the Supervisor template setup for identity configuration and verification.
Key Features
- **Persistent Knowledge**: Agents accumulate expertise across sessions (L-docs)
- **Shared Memory**: KB as collaboration substrate, not hidden AI state
- **Requirements-Driven**: Human-level requirements ground testable specifications (requirements/)
- **Gate Discipline**: Explicit decision points with human approval (GOVERNANCE_PRINCIPLES)
- **Fleet Coordination**: Multi-agent patterns with clear authority boundaries
- **Evidence-First**: Audit before architecture. Validate before shipping.
Supported Platforms
| Platform | Status | Integration |
|----------|--------|-------------|
| **Claude Code** | Baseline | CLAUDE.md, .claude/ |
| **Codex CLI** | Compatible | AGENTS.md, .codex/ |
| **Gemini CLI** | Compatible | AGENTS.md |
| Cursor | Experimental | .cursor/rules |
| Aider | Experimental | CONVENTIONS.md |
See CLI Support Matrix for details.
What AGET is NOT
- **Not an AI model or runtime.** AGET is a governance layer that works with any LLM backend.
- **Not a replacement for Claude Code or Codex CLI.** AGET sits above these platforms.
- **Not an autonomous system.** AGET requires human supervision and gate discipline.
- **Not coding-only.** AGET supports advisory, consulting, research, and general knowledge work.
Templates
12 core archetypes across 13 registered templates, including the specialized document-processor variant:
| Template | Use Case |
|----------|----------|
| **template-supervisor-aget** | Fleet coordination (**start here**) |
| template-worker-aget | Task execution |
| template-developer-aget | Development workflows |
| template-advisor-aget | Advisory with personas |
| template-analyst-aget | Data analysis |
| template-architect-aget | System design |
| template-researcher-aget | Research workflows |
The Problem
Your CLI agents lose context between sessions. Knowledge resets daily. Agents in the same fleet can't learn from each other. You have no confidence that what worked yesterday still works today.
AGET fixes this. It gives your agents persistent knowledge, shared memory, fleet coordination, and human-supervised governance. Your agents accumulate domain expertise that compounds across sessions.
**New to AGET?** Listen to the audio introduction — a narrated walkthrough of session rhythm, key skills, and fleet creation.
v3.36.0 release candidate
See the release notes and migration handoff. Availability and receiver acceptance require separate verification.
The migration kit's test-body witness requires Python 3.12+ (sys.monitoring); on Python 3.10/3.11 a run needing that witness returns INCONCLUSIVE by design (weekly-train:R15).
Quick Start
Start with the Supervisor template. It coordinates your fleet and can create new agents.
See GETTING_STARTED.md for archetype selection and setup guidance. Follow the Supervisor template setup for identity configuration and verification.
Key Features
- **Persistent Knowledge**: Agents accumulate expertise across sessions (L-docs)
- **Shared Memory**: KB as collaboration substrate, not hidden AI state
- **Requirements-Driven**: Human-level requirements ground testable specifications (requirements/)
- **Gate Discipline**: Explicit decision points with human approval (GOVERNANCE_PRINCIPLES)
- **Fleet Coordination**: Multi-agent patterns with clear authority boundaries
- **Evidence-First**: Audit before architecture. Validate before shipping.
Supported Platforms
| Platform | Status | Integration |
|----------|--------|-------------|
| **Claude Code** | Baseline | CLAUDE.md, .claude/ |
| **Codex CLI** | Compatible | AGENTS.md, .codex/ |
| **Gemini CLI** | Compatible | AGENTS.md |
| Cursor | Experimental | .cursor/rules |
| Aider | Experimental | CONVENTIONS.md |
See CLI Support Matrix for details.
What AGET is NOT
- **Not an AI model or runtime.** AGET is a governance layer that works with any LLM backend.
- **Not a replacement for Claude Code or Codex CLI.** AGET sits above these platforms.
- **Not an autonomous system.** AGET requires human supervision and gate discipline.
- **Not coding-only.** AGET supports advisory, consulting, research, and general knowledge work.
Templates
12 core archetypes across 13 registered templates, including the specialized document-processor variant:
| Template | Use Case |
|----------|----------|
| **template-supervisor-aget** | Fleet coordination (**start here**) |
| template-worker-aget | Task execution |
| template-developer-aget | Development workflows |
| template-advisor-aget | Advisory with personas |
| template-analyst-aget | Data analysis |
| template-architect-aget | System design |
| template-researcher-aget | Research workflows |
- 来源
- github · https://github.com/aget-framework/aget
- 按领域浏览
- 开源项目
- Traction
- 12 ★
- 收录于
- 7 天前
更多开源项目
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。