86 lines
2.0 KiB
Markdown
86 lines
2.0 KiB
Markdown
# TOOLS.md — 套件开发工具与资源
|
||
|
||
## 安装
|
||
|
||
### agentgis-cli(命令行工具)
|
||
|
||
```bash
|
||
pip install https://git.mercator.cn/api/packages/SuiteHub/generic/agentgis-cli/0.1.0/agentgis_cli-0.1.0-py3-none-any.whl
|
||
|
||
# 查看版本
|
||
agc --version
|
||
|
||
# 配置 API Key
|
||
agc config set api-key mk_xxxxxxxxxxx
|
||
```
|
||
|
||
### agentgis-sdk(Python 开发包)
|
||
|
||
```bash
|
||
pip install https://git.mercator.cn/api/packages/SuiteHub/generic/agentgis-sdk/0.1.0/agentgis_sdk-0.1.0-py3-none-any.whl
|
||
```
|
||
|
||
### gis-actions(本地执行器,可选)
|
||
|
||
开发者如需本地测试套件执行:
|
||
|
||
```bash
|
||
# 安装 gis-actions 软件包
|
||
# Debian/Ubuntu:
|
||
wget -O agentgis-worker.deb https://git.mercator.cn/api/packages/SuiteHub/generic/gis-actions/v2.0.0-alpha/agentgis-worker_2.0.0-alpha_all.deb
|
||
sudo dpkg -i agentgis-worker.deb
|
||
# 配置后启动
|
||
sudo systemctl start agentgis-worker
|
||
```
|
||
|
||
---
|
||
|
||
## CLI 用法
|
||
|
||
```bash
|
||
# 🔍 查市场(写代码前先做这个)
|
||
agc suites search <关键词> # 搜索已有套件
|
||
agc suites list # 列出所有套件
|
||
|
||
# 初始化新套件
|
||
agc init ./my-analysis
|
||
|
||
# 测试执行
|
||
agc run <suite-id> --inputs '{"expression": "1+1"}'
|
||
|
||
# 发布到市场
|
||
agc publish ./my-analysis
|
||
```
|
||
|
||
## API 端点
|
||
|
||
| 用途 | 端点 |
|
||
|------|------|
|
||
| 发布 Suite | `POST https://suites.mercator.cn/api/v1/suites` |
|
||
| 执行任务 | `POST https://suites.mercator.cn/api/v1/task` |
|
||
| 数据类型列表 | `GET https://suites.mercator.cn/api/v1/data-types` |
|
||
| 业务类型列表 | `GET https://suites.mercator.cn/api/v1/business-types` |
|
||
| API 文档 | `https://suites.mercator.cn/docs` |
|
||
|
||
## 套件结构
|
||
|
||
```
|
||
my-suite/
|
||
├── workflow.yaml # 步骤定义(必填)
|
||
└── scripts/
|
||
└── run.py # 执行入口脚本
|
||
```
|
||
|
||
## 基础镜像
|
||
|
||
所有套件在 gis-base 镜像中执行,包含:Python 3, GDAL, Shapely, GeoPandas, numpy
|
||
|
||
## 知识库
|
||
|
||
- `knowledge/script-data-types.md` — 脚本数据类型分类规范
|
||
- `knowledge/script-dependencies.md` — 脚本依赖管理规范
|
||
|
||
## 文档
|
||
|
||
公开文档在 `SuiteHub/agent-profiles` 仓库中。
|