42ADocs
MCP

MCP overview

What the @42a/mcp server exposes.

Once installed, you can ask Claude things like:

"What was my visibility score last week, and which pages picked up the most citations?"

…and Claude will call get_org_vitals and list_page_vitals on your behalf, then summarize the answer.

What's exposed

ToolEndpoint
list_organizations, get_me/api/v1/organizations, /api/v1/me
list_projects, get_project/api/v1/projects
list_top_brands_in_project/api/v1/projects/{id}/top-brands
list_top_prompts_in_project/api/v1/projects/{id}/top-prompts
list_top_websites_in_project/api/v1/projects/{id}/top-websites
list_prompts, get_prompt/api/v1/prompts
list_top_brands_in_prompt/api/v1/prompts/{id}/top-brands
list_topics/api/v1/topics
list_top_brands_in_topic/api/v1/topics/{id}/top-brands
list_top_websites_in_topic/api/v1/topics/{id}/top-websites
list_brands, get_brand/api/v1/brands
get_brand_stats/api/v1/brands/{id}/stats
list_citations/api/v1/citations
get_org_vitals/api/v1/vitals
get_category_vitals/api/v1/vitals/by-category
list_top_websites_in_vertical/api/v1/categories/{category}/top-websites
list_general_vitals/api/v1/general-vitals
list_page_vitals/api/v1/page-vitals

Sample prompts they cover:

  • "How did my brand visibility improve in the last week?"get_brand_stats(from: "2026-05-08", to: "2026-05-14") (look at trend.score_change).
  • "How many times was my brand mentioned?"get_brand_stats (mention_count).
  • "What are the top X mentioned brands in project Y?"list_top_brands_in_project.
  • "Which prompts drove the most mentions in project Y?"list_top_prompts_in_project.
  • "Top brands mentioned in prompt X?"list_top_brands_in_prompt.
  • "Who is the best mentioned brand in topic X?"list_topics -> list_top_brands_in_topic.
  • "Top cited domains in topic X?"list_topics -> list_top_websites_in_topic.
  • "What are the top cited websites in project Y in the past week?"list_top_websites_in_project(from: "2026-05-08", to: "2026-05-14").
  • "Top cited domains in the comparison vertical?"list_top_websites_in_vertical(category: "comparison").
  • "What should I improve on my website?"list_general_vitals(brand_id) for site-wide config (robots.txt, llms.txt, sitemap, …) then list_page_vitals(sort: "score_low", limit: 5) to drill into per-page issues.
  • "What is my strongest vertical?"get_category_vitals (first item).

The schemas mirror the REST API one-to-one - agents that have read the OpenAPI spec already know how to call these tools.

Annotations & structured output

Every tool declares the same MCP annotations, and they hold for the whole toolset:

  • readOnlyHint: true — every tool is an HTTP GET against the 42A API; nothing is ever created, modified, or deleted.
  • destructiveHint: false — read-only retrieval; there is no state to destroy or overwrite.
  • idempotentHint: true — the same arguments return the same result within a stable time window.
  • openWorldHint: false — tools query a closed, tenant-scoped analytics dataset behind a fixed API; they never reach arbitrary external systems or the open web.

Every tool also declares an outputSchema, and responses include structuredContent matching it. All payloads share the envelope { data, next_cursor, meta }, so clients and models can pick response fields by name instead of parsing free text.

Next: install it.

Last updated