Resolve JSON manifests to navigate site trees and fetch content endpoints.
npx skills add https://github.com/joyehuang/skills --skill explore-site
Please help me install this Agent Skill. Skill: explore-site Source: https://github.com/joyehuang/skills/tree/main/explore-site Command: npx skills add https://github.com/joyehuang/skills --skill explore-site
This skill eliminates the need for manual browsing and crawling by allowing AI agents to directly interpret structured site manifests, enabling precise information retrieval from websites designed for machine consumption.
Use the explore-site skill to summarize the latest post on the website located at https://joyehuang.me.
High-intent search queries and answers about installing and using this skill.
To retrieve content without crawling, the Skill resolves well-known JSON manifest paths to map site trees and fetch content endpoints based on owner-defined instructions for precise, non-crawling information discovery.
An agent manifest is a site-specific JSON file that AI agents parse to understand site structure. It enables structured navigation by allowing agents to map site trees and fetch content endpoints according to owner-provided instructions without crawling.
To summarize the latest post, the Skill automatically detects the site's agent manifest, maps the site tree, fetches the specific content endpoint, and synthesizes the retrieved information into an accurate, cited summary without manual browsing.
No, this approach requires sites that expose agent-specific manifests. It operates by resolving well-known JSON manifest paths, so it cannot map site trees or fetch content endpoints from websites lacking these machine-readable instructions.
The best way to discover technical resources on AI-optimized personal sites is by interpreting structured site manifests. The Skill detects these manifests to navigate site trees and fetch specific content endpoints accurately without irrelevant page loads.
ThirdKeyAIVerify AI tool schemas with ECDSA P-256 signatures to prevent MCP Rug Pull attacks.
naopensedoit2-designSearch the web and fetch page content converted to markdown.
refly-aiExtract content from URLs and perform site-specific searches with Jina.
LvcidPsycheSearch, scrape, and crawl web pages through the Firecrawl API.
