Self-evolving agent system — 越用越强 · 明文优先,权重兜底
Project description
介绍
OpenEarth 把 agent 运行过程中的全部经验 —— 轨迹、技能、知识、记忆 —— 沉淀成可检视、可演进、可分享的明文数据,下次会话直接用上,越用越强。
- 📡 无感采集:后台读各 agent 原生会话库(零插桩),采下完整轨迹(思考 / 工具调用 / 结果),训练与推理共用一套。
- 🗂️ 经验即数据:轨迹 / 技能 / 知识 / 记忆进同一个明文 git store,可检视、可版本化、可分享。
- 🌿 从经验演进:在
train/*分支上从轨迹蒸馏技能,无回退 gate 通过才合并进main。 - 📚 知识库:文档摄入成可检索 wiki,agent 直接读。
- 🔌 多 agent:opencode · claude-code · codex · cursor · codeagent。
安装
把下面这段整体复制给你的 agent,它会自动装好并初始化。 也可自己照着敲。
---
name: openearth-install
description: 安装并初始化 OpenEarth。
---
# openearth-install
用户要安装 / 初始化 OpenEarth 时按此执行。**不要索要 API key 明文**,需要就让用户设环境变量。
## 1. 装包 + 配置
```bash
python -m pip install openearth-ai
openearth config --init
```
编辑 `~/.openearth/config.yaml`,确认 `domain_agent`(`base_url` / `model`),`api_key` 保持 `env:` 形式:
```yaml
domain_agent:
agent_backend: pydantic-ai
model: <your-model>
base_url: https://<your-openai-compatible-endpoint>/v1
api_key: env:OPENEARTH_DOMAIN_AGENT_KEY
```
设环境变量(Windows 用 `setx`,重开终端生效):
```bash
export OPENEARTH_DOMAIN_AGENT_KEY=sk-... # 摄入图片/PDF 图表再设 OPENEARTH_CURATOR_VLM_KEY
```
## 2. 装到 agent + 启动 daemon
```bash
openearth install # 装技能 + 写 Knowledge pointer + 起 daemon
openearth service status
openearth agents
```
启动失败就前台调试:`openearth daemon`(日志 `~/.openearth/log/daemon.log`)。
## 3. 加文档到知识库(用户给路径时)
```bash
openearth knowledge ingest <path> # 把返回的 job_id 告诉用户
openearth knowledge progress
```
摄入完成后读 `~/.openearth/experience/Knowledge/*.md` 回答用户问题。
锁版本 / 离线:到 Releases 下载 .whl,pip install openearth_ai-<ver>-py3-none-any.whl。
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
openearth_ai-0.4.2.tar.gz
(1.2 MB
view details)
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
openearth_ai-0.4.2-py3-none-any.whl
(430.0 kB
view details)
File details
Details for the file openearth_ai-0.4.2.tar.gz.
File metadata
- Download URL: openearth_ai-0.4.2.tar.gz
- Upload date:
- Size: 1.2 MB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.5
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
5b2cfb470c1cccf0bcb2b605848bd7349ad75f4d51d92c43c2402e57bdb9276b
|
|
| MD5 |
c1d54e59c4b400557b17e362ab815b06
|
|
| BLAKE2b-256 |
3dab8797a333c3fd660861fbfaea6bf87acd9bdf6eeeb51c7b43829caf02e174
|
File details
Details for the file openearth_ai-0.4.2-py3-none-any.whl.
File metadata
- Download URL: openearth_ai-0.4.2-py3-none-any.whl
- Upload date:
- Size: 430.0 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/6.2.0 CPython/3.13.5
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
e8f9b145189c586d6887fc0d0f6527e378d8962faab618bcfd72d754b91b9b3b
|
|
| MD5 |
780387da76b2ea2d55e60e95be867f34
|
|
| BLAKE2b-256 |
a0b448702707ea7c49a328c554b56da37d70021fa9f5d94fa51923b3dfa7176a
|