Every read here is scoped to the public index — the brand brief, the prompt records and the index statistics each carry that word in their own descriptions — so this is a research surface rather than an account dashboard, and nothing in it is tied to owning the brand you are asking about. The useful shape is that parse_get_brand does not stop at a score: it hands back the prompts a brand surfaces in, the sources cited around it, and a list of next research questions, which is usually enough to decide where the second call should go. search and fetch exist for clients that insist on those names and resolve to the same data.
A read-only front end to the public Parse index of brands, organic AI prompts and citation sources, with six tools including generic search and fetch aliases for clients that expect those names.
- One entry point for research: parse_search finds brands, organic AI prompts, citation sources and market niches, takes a types filter and a limit, and its description names it the first call when someone mentions a brand, category, source or AI-visibility question.
- A brand brief in a single call: parse_get_brand returns the Parse score, strengths, weak spots, top prompts, citation sources, related brands and suggested next research questions for a slug or id.
- The exact question behind a result: parse_get_prompt returns one public organic prompt by slug, so the AI-search question itself can be read rather than inferred.
- Index scale and freshness: parse_get_stats reports the tracked brands, organic prompts and citation observations behind the answers.
- Compatibility aliases: search mirrors parse_search for clients that expect a generic search tool, and fetch resolves prefixed ids such as brand:stripe or prompt:best-crm into JSON-text results with human-readable text.
No credentials are configured on the server, and the index it reads is the public one.
One command — npx -y -p @parse-gl/mcp parse-mcp