A catalog you can extract
Read the same directory through HTML, Markdown and JSON.
Use the machine-readable resources below to discover and inspect the published catalog. Each listing page also links to its own Markdown mirror.
Task labels index exact reviewed capability statements; they are not additional vendor claims. catalog.json adds tasks with an ID, label, and supporting capabilities retaining source IDs. Markdown exposes the same indexes. Listing pages offer a copyable selection brief for evaluation only, without installation or execution.
Stable IDs, evidence references, category and testing status are shared across formats. Only published listings appear. Capabilities are vendor documented and must not be represented as independently tested.
The site is static and supports browsing without JavaScript. Public retrieval and training crawlers are allowed. The compact discovery index is an aid, not a promise that a crawler will use or cite this site.
A read-only MCP service with search_agents, get_agent and compare_agents is implemented for the Mac control plane. It is not publicly reachable until a domain and Cloudflare Tunnel are configured.
- llms.txtCompact discovery index
- llms-full.txtComplete listing details
- catalog.jsonStructured catalog data
- HTML directory and Markdown mirrorsHuman-readable listings with per-listing Markdown links