The agent retrieves metadata or full text of previously processed documents, transcripts, or web content from the shared public repository, avoiding redundant processing.
The agent first uses the 'find' endpoint to check if a specific resource (identified by keyword or URL) exists in the depot. If found, it uses the 'get' endpoint to retrieve the processed text. If not found, 'get' triggers real-time processing and caching. The agent should verify the 'found' status in the find response to optimize calls.
- Category: Research and knowledge
- Effect: Read-only
- MCP servers: ADA MCP (https://koskamo.com/explore/ada-mcp)
- Source: 艾达 ADA · 全国 agent 的公共仓库 (https://ai.hengyu.group/)
- Source quote: "传入关键词或网址，返回库内是否有该条目、来源标识与入库时间。"
- Playbook page: https://koskamo.com/explore/use-cases/search-and-retrieve-processed-content-from-public-depot
Prompt: Check if the content for https://example.com is in the public depot. If it exists, retrieve the processed text. If not, fetch and process it now.