Skip to content
AnswerixAI

MCP server

Connect Claude, Cursor and other MCP clients to the AnswerixAI MCP server with an API key, and query your AI visibility data with 24 tools.

Updated

AnswerixAI runs a Model Context Protocol (MCP) server, so AI assistants can read your AI visibility data while you chat with them. Ask your assistant "how is my brand doing on ChatGPT this week?" and it calls AnswerixAI's tools and answers with your numbers. There's nothing to install on the AnswerixAI side: you need the MCP endpoint and an API key. To chat with your data inside the dashboard instead, use Chat.

What do I need to connect?

To connect an AI assistant to the AnswerixAI MCP server you need two things, both from Settings → API Keys: the MCP endpoint URL, which ends in /api/mcp, and an ans_ API key.

  1. The MCP endpoint. Open Settings and select the API Keys tab. Copy the URL under MCP endpoint. It's the address you use to sign in to AnswerixAI followed by /api/mcp, for example https://answerixai.com/api/mcp. The server also answers at /mcp on the same address.
  2. An API key. On the same tab, click New key, give the key a name (for example "MCP — my laptop") and click Create. Copy the key, which starts with ans_. It's shown only once, so store it somewhere safe.

The examples below use https://answerixai.com/api/mcp. If the MCP endpoint shown in your Settings is different, use yours instead.

How does the MCP server authenticate?

The AnswerixAI MCP server uses the MCP Streamable HTTP transport and authenticates every request with an ans_ API key sent as a bearer token. It doesn't support OAuth sign-in or your dashboard login. Send the key in this header:

Authorization: Bearer ans_your_key_here

The server doesn't support OAuth sign-in, and it doesn't accept your dashboard login. Requests without a valid ans_ key get a 401 error. An API key sees the brands in its creator's organization. See Authentication.

How do I connect Claude Code?

To connect Claude Code to the AnswerixAI MCP server, run claude mcp add with the HTTP transport, the MCP endpoint and your API key in an Authorization header. Replace the key with your own:

claude mcp add --transport http answerix https://answerixai.com/api/mcp \
  --header "Authorization: Bearer ans_your_key_here"

Then ask Claude "List my AnswerixAI brands". Claude calls list_brands and shows your brands.

How do I connect Claude Desktop?

Claude Desktop's config file only starts local servers, and its Connectors settings use OAuth, which the AnswerixAI server doesn't offer. Use mcp-remote, a small bridge that runs on your computer and forwards requests to AnswerixAI with your key. You need Node.js 18 or later.

Edit claude_desktop_config.json: ~/Library/Application Support/Claude/claude_desktop_config.json on macOS, or %APPDATA%\Claude\claude_desktop_config.json on Windows:

{
  "mcpServers": {
    "answerix": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://answerixai.com/api/mcp",
        "--header",
        "Authorization:${AUTH_HEADER}",
        "--transport",
        "http-only"
      ],
      "env": {
        "AUTH_HEADER": "Bearer ans_your_key_here"
      }
    }
  }
}

The key goes in env, with no spaces around the : in args, because Claude Desktop on Windows doesn't escape spaces inside args. This form works on macOS too. Restart Claude Desktop, and the answerix tools appear in the tools menu.

How do I connect Cursor?

To connect Cursor to the AnswerixAI MCP server, add the MCP endpoint URL and an Authorization header with your API key to .cursor/mcp.json in your project, or to ~/.cursor/mcp.json for all projects:

{
  "mcpServers": {
    "answerix": {
      "url": "https://answerixai.com/api/mcp",
      "headers": {
        "Authorization": "Bearer ans_your_key_here"
      }
    }
  }
}

Can I use other MCP clients?

Yes. Any MCP client that supports the Streamable HTTP transport and lets you set a custom Authorization header can connect with the endpoint and key. For clients that only run local servers, use mcp-remote as shown for Claude Desktop.

Which tools does the MCP server provide?

The server provides 24 tools. Start with list_brands to get a brand_id, then pass it to the other tools. Dates are ISO timestamps, such as 2026-09-01T00:00:00Z. Arguments marked with * are required.

Which tools cover brands and visibility?

Five AnswerixAI MCP tools cover brands and visibility: list_brands, get_visibility_summary, get_visibility_trend, get_competitor_comparison and get_prompt_performance. Every tool except list_brands requires brand_id.

Tool Arguments What it returns
list_brands none The brands you can access, with id, name, slug, industry, region and creation date.
get_visibility_summary brand_id*, date_from, date_to, model, region The headline AI Visibility score (0–100) with its components, total mentions and citations, and the top 5 competitors by mentions.
get_visibility_trend brand_id*, date_from, date_to, granularity (day or week), model, region, topic_id Visibility over time in day or week buckets (weeks start on Monday). Each bucket has the result count, mentions, citations and average competitor score. The headline ai_visibility_score is filled in for day buckets only.
get_competitor_comparison brand_id*, date_from, date_to, model, region, topic_id Your brand and every tracked competitor with their AI Visibility score, mentions, citations and appearances, plus share of voice overall and per platform. It always compares against all tracked competitors, so there's no competitor argument.
get_prompt_performance brand_id*, date_from, date_to, topic_id, sort_by (visibility, mentions, citations, appearances), order (desc, asc), limit (1–100, default 10), model, region Visibility, mentions, citations, appearances and competitor score for each prompt, totalled across platforms. There's no per-platform breakdown.

Which tools cover topics and prompts?

Six AnswerixAI MCP tools cover topics and prompts: list_topics, list_topic_suggestions, accept_topic_suggestion, dismiss_topic_suggestion, list_prompts and get_prompt_volumes. New topic suggestions and new volume analyses can't be started over MCP.

Tool Arguments What it returns
list_topics brand_id* Your topics, each with its number of prompts.
list_topic_suggestions brand_id* Pending AI-generated topic suggestions. New suggestions can't be generated over MCP.
accept_topic_suggestion suggestion_id* Accepts a suggestion and creates exactly one topic from it.
dismiss_topic_suggestion suggestion_id* Dismisses a suggestion so it doesn't come back.
list_prompts brand_id*, topic_id, is_active, limit (1–500, default 100) Your tracked prompts, newest first, with topic, platforms, models, regions and whether each one is active.
get_prompt_volumes brand_id*, limit (1–500, default 100) Search demand for prompts that already have a volume analysis: keywords, Google search volume, estimated AI search volume and competition. New analyses can't be started over MCP.

Which tools cover citations, traffic and shopping?

Four AnswerixAI MCP tools cover citations, AI traffic and shopping: list_citations, get_ai_traffic, list_shopping_cards and get_product_visibility. Each one requires brand_id.

Tool Arguments What it returns
list_citations brand_id*, date_from, date_to, model, region, topic_id, limit (1–200, default 50), source_filter (all, owned, competitor, external) The domains and URLs AI engines cite, with a source-type breakdown and counts.
get_ai_traffic brand_id*, date_from, date_to Visits from AI engines: total, per platform, top landing pages and countries.
list_shopping_cards brand_id*, role (own, competitor, other), platform, region, limit (1–200, default 50), cursor Product cards AI engines showed in shopping answers, one page at a time.
get_product_visibility brand_id, product_title Appearances, platforms, regions, dates, rating, reviews and merchants for one product, using a title from list_shopping_cards.

Which tools cover content opportunities?

Five AnswerixAI MCP tools cover content opportunities: list_content_opportunities, get_content_opportunity, generate_content_brief, generate_content_draft and update_opportunity_status. Generating a brief or a draft uses your plan's quota; reading an existing brief doesn't.

Tool Arguments What it returns
list_content_opportunities brand_id*, status (new, sent, in_progress, done, dismissed), impact (high, medium, low), type (owned, earned), limit (1–200, default 50) Content opportunities, highest opportunity score first.
get_content_opportunity opportunity_id* Full details of one opportunity, including its content brief if one exists. Reading a brief this way doesn't use quota.
generate_content_brief opportunity_id* Generates a new content brief. It always generates a fresh brief, replacing any existing one, and uses your brief quota.
generate_content_draft opportunity_id* Generates a full article draft from the opportunity's brief, with meta title, meta description, slug, FAQs, key takeaways and structured data. The opportunity needs a brief first. It always replaces any existing draft and uses your content draft quota.
update_opportunity_status opportunity_id, status (new, sent, in_progress, done, dismissed) Moves an opportunity to a new status.

Which tools cover site audits?

Four AnswerixAI MCP tools cover site audits: run_site_audit, get_site_audit, list_site_audits and get_site_audit_quota. Only run_site_audit uses a Site Audit credit.

Tool Arguments What it returns
run_site_audit brand_id, url Starts a Site Audit of the page and returns its id with the status running. Uses one Site Audit credit.
get_site_audit audit_id* An audit's status, total score, category scores, signals and fix recommendations. Poll it until the status is completed or failed.
list_site_audits brand_id* Your recent audits, newest first.
get_site_audit_quota none Your monthly Site Audit allowance: used, limit and remaining. A limit of -1 means unlimited.

For the model argument, pass a model name or a comma-separated list. For region, pass a region code such as US.

Which tools use credits?

Three AnswerixAI MCP tools use your plan's quotas, the same quotas the dashboard uses: run_site_audit, generate_content_brief and generate_content_draft. Every other tool is free to call:

  • run_site_audit uses one Site Audit credit.
  • generate_content_brief uses one brief generation.
  • generate_content_draft uses one content draft.

Every other tool only reads data, except accept_topic_suggestion, dismiss_topic_suggestion and update_opportunity_status, which change data but don't use credits. The server tells AI assistants to use the credit-consuming tools only when you ask. If a quota has run out, the tool returns the limit message instead. See Billing and plans for each plan's quotas.

What can I ask?

Once the AnswerixAI MCP server is connected, you can ask your AI assistant about your brands, visibility score, competitors and share of voice, cited URLs and topics, or ask it to run a site audit. Some examples:

  • "List my AnswerixAI brands."
  • "What's my visibility score for the last 7 days?"
  • "Who are my top competitors by share of voice this month?"
  • "Which of our own URLs did AI engines cite this month?"
  • "Which topics have the fewest prompts?"
  • "Run a site audit on our pricing page and summarize the weakest category."

Is there a REST version of these tools?

Yes. Each tool also has a REST endpoint under /api/mcp/* that takes the same ans_ key, which is handy for scripts. See the REST API reference.

How do I revoke access?

Open Settings, select the API Keys tab, click the trash icon next to the key and confirm with Revoke. Clients using that key lose access immediately. Create a separate key for each client, so you can revoke one without breaking the others.

Related

Was this page helpful?