Back to Details

hence-search

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Hence Search

Hence 搜索

Install from your project root directory. Running 
npx skills add
from a subdirectory will install into the wrong location.
Search the Hence gallery to find projects and draw inspiration from what others are building with AI.
请从项目根目录安装。 从子目录运行
npx skills add
会安装到错误的位置。
搜索Hence画廊，查找他人用AI构建的项目以获取灵感。

Workflow

工作流程

0. Authenticate

0. 身份验证

Check for existing credentials. If none found, start the device flow:
bash
python scripts/auth.py --check
If the check fails, run the device flow:
bash
python scripts/auth.py
The script will print a URL and a one-time code. Before running the command, tell the user they'll need to:
  1. Open the URL in their browser
  2. Log in to Hence if they aren't already (via GitHub or Google)
  3. Enter the code shown in the terminal
The script will wait and automatically complete once the user approves. No further action is needed from the agent after that.
For CI/CD environments, pass an API key directly:
bash
python scripts/auth.py <api-key>
检查现有凭据。如果未找到，启动设备流程：
bash
python scripts/auth.py --check
如果检查失败，运行设备流程：
bash
python scripts/auth.py
脚本会打印一个URL和一次性验证码。在运行该命令前，请告知用户他们需要：
  1. 在浏览器中打开该URL
  2. 如果尚未登录，请登录Hence（通过GitHub或Google）
  3. 输入终端中显示的验证码
脚本会等待，用户批准后将自动完成流程。之后Agent无需执行其他操作。
对于CI/CD环境，直接传入API密钥：
bash
python scripts/auth.py <api-key>

1. Determine search strategy

1. 确定搜索策略

  • Specific query (e.g. "Rust CLI tools"): run 
    search.py
    directly with keywords.
  • Broad browsing (e.g. "show me cool stuff"): run 
    fetch_metadata.py topics
    first, suggest 3–5 relevant categories, then search by topic.
  • 特定查询（例如“Rust CLI工具”）：直接运行
    search.py
    并传入关键词。
  • 宽泛浏览（例如“给我展示有趣的内容”）：先运行
    fetch_metadata.py topics
    ，推荐3–5个相关分类，然后按主题搜索。

2. Execute search

2. 执行搜索

bash
python scripts/search.py "productivity cli"
python scripts/search.py "" --topic game --limit 10
python scripts/search.py "dashboard" --topic react --offset 20
For valid topic slugs:
bash
python scripts/fetch_metadata.py topics
Pass 
--json
to either script for raw JSON output when further processing is needed.
bash
python scripts/search.py "productivity cli"
python scripts/search.py "" --topic game --limit 10
python scripts/search.py "dashboard" --topic react --offset 20
如需获取有效的主题别名：
bash
python scripts/fetch_metadata.py topics
当需要进一步处理时，可向任一脚本传入
--json
参数以获取原始JSON输出。

3. Present results

3. 展示结果

For each project include:
  • Title and pitch — the name and one-liner
  • Built with — agent/model if available
  • Link — 
    https://hence.sh/p/<id>
每个项目需包含：
  • 标题与简介 — 项目名称和一句话描述
  • 构建工具 — 若有，标注使用的Agent/模型
  • 链接 — 
    https://hence.sh/p/<id>

4. Connect inspiration

4. 关联灵感

After showing results, ask: "Does any of this spark ideas for your current build?"
If the user identifies a project as inspiring, store its ID and title via 
save_memory
so the 
hence-share
skill can suggest an "Inspired by" link later.
展示结果后，询问：“这些内容是否为你当前的构建工作带来了灵感？”
如果用户认为某个项目很有灵感，通过
save_memory
存储其ID和标题，以便后续
hence-share
技能可以推荐“灵感来源”链接。

5. Paginate

5. 分页

If results are truncated, offer to load more by incrementing 
--offset
.
如果结果被截断，可通过递增
--offset
参数提供加载更多内容的选项。

API details

API详情

See references/api.md for full endpoint documentation, response schemas, and available metadata endpoints.
请查看references/api.md获取完整的端点文档、响应架构和可用的元数据端点。