Merekit Public Discovery API

This site exposes public discovery metadata for agents and crawlers. The operational surface is the mere CLI, 15 bundled app adapters, Business workspace and website lifecycle commands, the Dynasite provider surface, the standalone @merekit/link graph CLI, mere.run setup for local AI/runtime work, and the local stdio MCP server.

The full Mere trifecta is mere.business for cloud workspaces and durable approvals, merekit for the agent-readable command plane, and mere.run for local models, files, media, indexes, and generated outputs.

Endpoints

• GET /robots.txt crawl policy, AI crawler directives, Content Signals, and sitemap location.
• GET /sitemap.xml canonical public URLs.
• GET /waitlist Turnstile-protected CLI waitlist form.
• POST /api/waitlist sends a magic link after server-side Turnstile validation and rate limiting.
• GET /waitlist/confirm verifies the magic link and joins the waitlist.
• GET /api/package-versions npm latest dist-tags for public Merekit packages.
• GET /.well-known/api-catalog RFC 9727 linkset catalog.
• GET /.well-known/openapi.json OpenAPI description.
• GET /skills centralized Mere skill catalog.
• GET /skills/index.json R2-backed skill registry.
• GET /skills/{name}/SKILL.md skill artifact.
• GET /.well-known/agent-skills/index.json compatibility mirror for agent skill discovery.
• GET /.well-known/mcp/server-card.json local mere mcp serve server card.
• GET /.well-known/oauth-protected-resource and GET /.well-known/oauth-authorization-server OAuth discovery metadata.

Authentication

The discovery endpoints do not require authentication. The waitlist form requires a valid Turnstile token and an email magic-link click before it forwards to the mere.business capture endpoint. Product operations are delegated through app-owned CLI adapters; product secrets remain in their product-local stores.

Runtime setup

Local runtime work delegates to mere.run. Agents and operators can prepare that runtime with mere setup mere-run --json, then pull app-requested models with commands such as mere setup mere-run models --app media --json. Approved outputs can sync back into the cloud workspace.

Business and Dynasite lifecycle

mere business workspace ... is the normal workspace creation, status, setup, use, list, and handoff surface. Website operations stay workspace-aware through mere business site ...: existing-site import, net-new site requests, CMS/media edits, revisions, publishing, and static bundle upload/publish/rollback.

mere dynasite ... is the direct provider surface for diagnostics and recovery. Static bundles let agents publish client-side HTML/CSS/JS/assets through Dynasite without executing arbitrary server or Worker code.

• Business lifecycle docs: sawfwair.github.io/merekit-cli/business-site
• CLI command reference: sawfwair.github.io/merekit-cli/commands

MereKit Link

@merekit/link publishes the standalone mere-link binary for declaring work graphs in mere.link.yaml. The public @merekit/cli package bundles the same behavior as mere link .... Link validates surfaces, exports project context, plans Mere Projects sync records, evaluates operator policy, and can bridge to Executor-backed tools through declared policy and resource guards.

• Standalone package: @merekit/link
• Docs: sawfwair.github.io/merekit-link
• Source: github.com/sawfwair/merekit-link

Back to merekit