Skip to main content

Registry & Discovery

The registry is the public catalog of appgrammars. It supports full-text search, tag filtering, trending rankings, and AI-powered discovery features that surface related and complementary appgrammars.

Browsing the Registry​

The registry is accessible at /registry in the web UI or via GET /api/registry. Without any filters, it returns public appgrammars sorted by recency.

Add a ?q= query parameter to search by name and description:

GET /api/registry?q=kanban&limit=20&offset=0

Tag Filtering​

Appgrammars can have up to 3 tags from a controlled vocabulary of 14:

TagTagTag
saase-commerceinternal-tool
apiclidashboard
landing-pagemobilecms
crmmarketplacedevtool
ai-appgame

Filter by tag with ?tags=saas,api. Tags are set by the appgrammar author via PUT /api/appgrammars/:id/tags.

To get the full tag list programmatically: GET /api/tags.

GET /api/registry/trending returns appgrammars ranked by weighted community activity (stars, executions, and forks) over a rolling time window.

Optional query parameters: days (window size, default 30) and limit.

Discovery​

The registry provides three discovery endpoints per appgrammar, all computed with pure SQL (no inference):

GET /api/appgrammars/:id/related

Co-occurrence based: users who starred appgrammar X also starred Y. Returns appgrammars with the highest overlap in star audiences.

Similar Appgrammars​

GET /api/appgrammars/:id/similar

Finds appgrammars with structurally similar module compositions.

Breed Suggestions​

GET /api/appgrammars/:id/breed-suggestions

Returns appgrammars whose modules complement the target — low module overlap, likely to produce a richer child. These are good candidates for breeding.

Trust Badges​

Badges are computed at render time based on execution history:

BadgeCriteria
Verified100+ executions with 95%+ success rate
Popular500+ executions with 90%+ success rate

Badges appear on registry cards and the appgrammar detail page.

MCP Access​

The appgrammar_search MCP tool provides registry search for AI agents:

{
  "query": "e-commerce",
  "limit": 10
}

This returns matching appgrammars with slug, name, description, configuration, and stats.