Everything an AI agent needs to use FOSSHunter data, by design.

The catalog is machine-readable end to end: discovery documents, a public JSON API, an OpenAPI contract, MCP tools, and per-tool markdown exports — all read-only, rate limited, and citation-first.

Discovery

llms.txt and the full rulebook

Start with the concise discovery document; open the full rulebook for interpretation rules, the field dictionary, and the response contract before quoting metrics.

llms.txt llms-full.txt

HTTP API

A public JSON API with an OpenAPI contract

Search, filter, and compare published tools. 60 requests per minute per IP, no key required.

curl https://www.fosshunter.com/api/public/v1/tools?q=monitoring&sort=stars

Discovery endpoint OpenAPI document

MCP

Native MCP tools for agent runtimes

Connect any MCP client to the streamable HTTP endpoint. 30 requests per minute per IP.

{
  "mcpServers": {
    "fosshunter": {
      "url": "https://www.fosshunter.com/mcp"
    }
  }
}

Tools: search_tools · get_tool · compare_tools · find_alternatives · list_categories

MCP discovery

Exports

Markdown and JSON for every tool

Fetch citation-ready markdown or canonical JSON for any published tool, or the whole catalog as one markdown index. Responses carry ETags, canonical Link headers, and cache windows.

Catalog markdown Tool markdown template

Citation policy

Cite what you quote, including when it was observed

Repository metrics are point-in-time GitHub observations. When presenting FOSSHunter data, link the tool's canonical_url and include synced_at plus the catalog revision with any metric you quote. Abstain when a comparison returns insufficient_data.

How to use

Three supported flows

Search, then cite the HTML page. Call the search API, read the matching tool detail, and cite the tool HTML page as the source. When quoting repository metrics, get synced_at and the catalog revision from the detail API response or the per-tool export.
Read the exports. Fetch citation-ready Markdown or JSON per tool; these exports have canonical Link headers pointing to the tool HTML page. The whole-catalog Markdown index is a separate document whose canonical Link points to itself.
Connect an MCP client. Point any MCP-capable client at the streamable HTTP endpoint. MCP requires client-side configuration on your side and is a connection method, not a crawler grant.

Search API Tool markdown template MCP discovery

Crawler policy

Crawlable is not a promise of a citation

robots.txt allows all crawlers. Search indexing, training data collection, and fetching on a user's request are distinct activities governed by each crawler operator's own policies; FOSSHunter publishes no per-crawler training permissions. Machine-readable surfaces make citation possible — appearing in any search engine or AI answer is decided elsewhere, not promised here.

Get started

Open discovery About FOSSHunter