What is MCP (Model Context Protocol)?

MCP is an open standard, originally created by Anthropic, that lets AI applications connect securely to external tools and data. LLM Pulse runs a native MCP server, so assistants like ChatGPT and Claude can read your AI visibility data and act on it during a normal conversation.

Which AI clients can connect?

Any client that implements the MCP specification. That includes ChatGPT, Claude (desktop, web and mobile), Gemini, Cursor, Windsurf, Zed and your own custom agents. You add one URL and sign in once.

What data can I query through MCP?

Everything in your dashboard. Brand visibility and share of voice over time, per model breakdowns, mentions, citations and source domains, full AI answer text with sentiment, prompts and executions, recommendations, GEO Writer tasks and, on Scale and above, AI bot traffic and AI traffic from your web analytics. There are about 90 tools in total.

What actions can an agent take?

Beyond reading data, agents can add prompts and competitors, create and assign tags, annotate your timeline, brief or draft content with GEO Writer, regenerate recommendations and launch a technical GEO audit. Write actions need the write scope, and the costly ones ask you to confirm before they run.

Is it safe to let an AI assistant into my account?

Yes. Connections use OAuth 2.1, are scoped to your account, and you can revoke access at any time from your API keys page. Read and write are split into separate scopes, so you can grant read-only access, and expensive actions return a preview and wait for your confirmation.

Do I need an API key?

No. The OAuth flow ships with every client and every plan, so there is no token to copy or rotate. An API key is only needed for headless setups such as servers, cron jobs or custom backends, which are available on Scale and above.

Is MCP included in my plan?

Yes. MCP is available on every LLM Pulse plan, including the 14-day free trial, at no extra cost. Annotation tools are included on every plan. Some other tools retain their app plan gates, such as Agent Analytics on Scale and above.

How is MCP different from the REST API?

Same data, different interface. The REST API is for code you write and HTTP requests you make. MCP lets an AI agent reach the same metrics and actions conversationally, choosing the right tool on its own. Use whichever fits your workflow, or both.

Does LLM Pulse have an MCP server?

Yes. LLM Pulse provides a hosted MCP server that works with ChatGPT, Claude, Gemini, Cursor and any MCP-compatible client. It includes about 90 tools for reading data and taking action, connects through secure OAuth with no API key, and is available on every plan.

Available on every plan

Connect AI agents to your visibility data with MCP

Let ChatGPT, Claude, Gemini, Cursor and any other AI agent read your brand's AI visibility metrics and act on them through the Model Context Protocol. About 90 tools, free on every plan via OAuth, no API key required.

Start Free Trial

14-day free trial

Cancel anytime

Connect AI agents to your visibility data with MCP

Connect your AI client

Paste this URL into your AI client's MCP or connectors settings. The client opens a browser window, you sign in to LLM Pulse, and you are connected. Tokens refresh automatically.

ChatGPT / Claude / Gemini Add this URL as a custom connector in your AI client

https://api.llmpulse.ai/api/v1/mcp

Cursor / Windsurf / Zed Add to your IDE's MCP server settings

https://api.llmpulse.ai/api/v1/mcp

Headless (API key, Scale and above) Setup guide

npx mcp-remote https://api.llmpulse.ai/api/v1/mcp --header "Authorization: Bearer YOUR_API_KEY"

Connect to

MCP Endpoint

Features

Bring your AI visibility data into any AI workflow

MCP lets AI assistants read your LLM Pulse data and act on it in real time, with no manual exports. It is available on every plan, including the free trial.

Built for AI agents

MCP is the open standard for connecting AI applications to data and tools. LLM Pulse runs a native MCP server, so agents talk to it directly.

Ask in natural language

Query your visibility data by simply asking. Compare competitors, spot sentiment shifts or find your weakest model without opening a dashboard.

Always the latest numbers

Agents read the same fresh data as your dashboard. No stale CSV exports, no manual refresh, no copy and paste.

No API key required

Connect with OAuth 2.1. Sign in once and your assistant stays connected, tokens refresh on their own and you can revoke access anytime.

Read and write coverage

Query every metric in the platform and take action too, from adding prompts and competitors to drafting content and launching audits.

Works in every client

ChatGPT, Claude, Gemini, Cursor, Windsurf, Zed and custom agents all connect to the same endpoint with no special integration.

Data queries

Every metric your agents can read

These read tools expose the same data you see in the dashboard, structured so an AI assistant can pull exactly what it needs. Group by group, here is what an agent can ask for.

Projects and competitors

Discover what an account tracks before drilling into the numbers.

list_projects
Every project the connected user can access.
get_project_details
A project's industry, business model, primary products, audience, brand names, prompt counts per brand focus and data coverage (models and locales with data).
list_competitors
The competitors tracked in a project, with name and domain.
get_competitor_details
A competitor's matching names, mobile app IDs and icon URLs.

Tags, models and coverage

The dimensions you can slice any query by.

list_tags / list_collections
The tags (collections) you use to group prompts by topic or campaign.
list_models
AI models with data for a project, such as ChatGPT, Perplexity and Gemini.
list_locales
Countries and languages that have data, so a query can target the right market.

Visibility metrics over time

The core numbers that power your dashboard charts.

get_timeseries
Mentions, visibility, citations and sentiment day by day, for you and competitors.
get_summary
Total, min, max and latest value for any metric over a period.
get_sov
Share of Voice, your share of mentions versus competitors.
get_prompt_summary
Per-prompt mention rate, citation rate and average position.

AI model breakdown

How you perform across each AI model, not just overall.

get_ai_model_summary
Compare mentions, citations, sentiment and visibility across every model in a single call.
get_ai_model_position_distribution
Where you rank, position one, two, three and beyond, per model.
get_ai_overview_results
Google AI Overview availability over time, with a per-prompt breakdown.

Prompts and responses

The raw questions and the answers the models gave.

list_prompts
Every prompt with its type, locale and last run.
list_prompt_executions
Each run with model, timing, fan-out queries and whether it produced a mention.
list_answers
AI responses with prompt text, model and basic metrics.
get_answer
The full response text with its mentions, citations, sentiment and sources.

Mentions

Where your brand and competitors are named in answers.

list_mentions
Brand mentions for a project.
list_competitor_mentions
Competitor mentions, filterable by competitor and prompt.
list_all_mentions
Brand and competitor mentions together, each tagged with its actor.

Citations

The URLs the models cite in their answers.

list_citations
URLs cited in answers that mention your brand.
list_competitor_citations
URLs cited for your competitors.
list_all_citations
Brand and competitor citations together, each tagged with its actor.

Sources and citation intelligence

Deep analysis of which pages and domains the models trust.

list_sources
Every source URL the models cited, with domain and position.
get_top_sources
Domains ranked by responses or by average mention rate.
list_citation_groups
Citations grouped by URL, domain or host, with model breakdowns.
get_cited_url_details
URL-level citation intelligence with page-cache metadata.
list_cited_url_occurrences
Every execution where a given URL appeared.
get_cited_url_content
The cached page content plus the mention snippets for one URL.

Sentiment

How positively or negatively the models talk about you.

list_sentiments
The sentiment categories and how the data is structured.
list_detailed_sentiments
Sentiment records with comments, topics, scores and competitor context.

Recommendations

AI-generated actions to improve your visibility.

list_recommendations
Recommendation runs with type, status, summary and counts.
get_recommendation
Full detail for one run, including items and source references.

GEO Writer tasks

The status and output of content tasks an agent created.

list_intelligence_tasks
GEO Writer tasks for a project, filterable by type and status.
get_intelligence_task
One task with its status and finished result data.

Agent Analytics

Scale and above

How AI crawlers and assistants hit your own website.

get_agent_traffic
AI bot traffic per day, grouped by bot or by company.
list_agent_bots
The catalog of known AI bots and their parent companies.

AI traffic

Scale and above

Human visits arriving from AI assistants, measured from your connected web analytics.

get_ai_traffic
Users, sessions and conversions from ChatGPT, Perplexity, Gemini and more, grouped by AI source.

Actions

Actions your agents can take, not just data they read

MCP is not read-only. With the mcp:write scope an assistant can change your project the same way you would in the app, and the costly actions always ask before they run.

Add prompts in bulk

create_prompts

Add new tracked prompts to a project in one call. Respects your plan's prompt limit and skips duplicates automatically.

Add a competitor

create_competitor

Start tracking a new competitor across every model and locale. Validated against your plan's competitor limit.

Create a tag

create_collection

Create a tag to segment prompts by topic, funnel stage or campaign, and optionally attach prompts in the same call.

Tag prompts in bulk

assign_prompt_tags

Attach existing tags to many prompts at once, by id or by name. Idempotent, so re-running never creates duplicates.

Annotate the timeline

Growth and above create_annotation

Mark a date on your visibility charts, a campaign launch, product update or news event, so the whole team sees the context behind a change.

Brief and draft content

create_intelligence_task

Kick off a GEO Writer task, a brief, a full draft, a rewrite, PR angle analysis or a custom output, then poll it until it is ready.

Regenerate recommendations

launch_recommendations

Re-run AI recommendations for a project. Returns a preview and a confirmation token, so nothing is spent until you approve it.

Run a technical GEO audit

create_technical_geo_report

Launch the full technical GEO bundle for a URL, covering crawlability, schema, content readiness, robots.txt, agent readiness, llms.txt, discoverability and AI visibility.

Actions are safe by design

Write tools require the mcp:write OAuth scope, so a read-only connection can never change your data. Costly actions like regenerating recommendations or launching a technical audit return a preview first and wait for your explicit confirmation inside LLM Pulse.

MCP for your analytics stack

Connect a data source and query it next to your AI visibility, all through the same MCP endpoint.

Google Search Console MCP

Read clicks, impressions, top queries and pages, and tie them to your AI visibility.

Learn more

Google Analytics 4 MCP

Measure AI traffic with GA4 and correlate it with mentions, citations and share of voice.

Learn more

Connect in seconds, not minutes

Connect your AI client to LLM Pulse with a single OAuth flow. No token to copy, no config file to edit.

1

Copy your MCP URL

Your endpoint is https://api.llmpulse.ai/api/v1/mcp, the same for every customer. No API key to generate.
2

Add it to your AI client

Paste the URL into the connector or MCP settings of ChatGPT, Claude, Gemini, Cursor, Windsurf or any other MCP client.
3

Sign in once with OAuth

Your client opens a browser window. Sign in to LLM Pulse, approve access and you are connected. Tokens refresh automatically.
4

Query and act conversationally

Ask about visibility, competitors, sentiment, citations and recommendations, or have the agent add prompts and draft content for you.

What you can do with MCP

From a quick question to a fully automated workflow, MCP turns your AI visibility data into something any assistant can read and act on.

Ad-hoc analysis

Ask your assistant to analyze visibility trends, compare models or surface citation patterns, all without touching the dashboard.

Automated reporting

Have an agent assemble weekly visibility summaries, competitor breakdowns or an executive brief straight from your live data.

Intelligent monitoring

Build agents that watch your metrics and flag you when visibility drops, sentiment turns or a competitor gains ground.

Strategy and content

Let AI read your data and suggest content opportunities, then brief or draft the piece with GEO Writer in the same conversation.

Hands-free maintenance

Add prompts, competitors and tags, annotate launches and regenerate recommendations just by asking, instead of clicking through forms.

Custom agents and workflows

Wire LLM Pulse into your own agents, internal tools or orchestration so AI visibility data flows wherever your team works.

"LLM Pulse has allowed us to better organize our work and prioritize the actions that matter most. It gives us much greater clarity when implementing new ideas, and it makes reporting far easier, with metrics that were previously complex and difficult to obtain. Today it is part of how we plan and measure our visibility in AI search."

Sandra Vargas

Euroinnova

Frequently Asked Questions

Everything you need to know about connecting AI agents to LLM Pulse over MCP.

: MCP is an open standard, originally created by Anthropic, that lets AI applications connect securely to external tools and data. LLM Pulse runs a native MCP server, so assistants like ChatGPT and Claude can read your AI visibility data and act on it during a normal conversation.
: Any client that implements the MCP specification. That includes ChatGPT, Claude (desktop, web and mobile), Gemini, Cursor, Windsurf, Zed and your own custom agents. You add one URL and sign in once.
: Everything in your dashboard. Brand visibility and share of voice over time, per model breakdowns, mentions, citations and source domains, full AI answer text with sentiment, prompts and executions, recommendations, GEO Writer tasks and, on Scale and above, AI bot traffic and AI traffic from your web analytics. There are about 90 tools in total.
: Beyond reading data, agents can add prompts and competitors, create and assign tags, annotate your timeline, brief or draft content with GEO Writer, regenerate recommendations and launch a technical GEO audit. Write actions need the write scope, and the costly ones ask you to confirm before they run.
: Yes. Connections use OAuth 2.1, are scoped to your account, and you can revoke access at any time from your API keys page. Read and write are split into separate scopes, so you can grant read-only access, and expensive actions return a preview and wait for your confirmation.
: No. The OAuth flow ships with every client and every plan, so there is no token to copy or rotate. An API key is only needed for headless setups such as servers, cron jobs or custom backends, which are available on Scale and above.
: Yes. MCP is available on every LLM Pulse plan, including the 14-day free trial, at no extra cost. Annotation tools are included on every plan. Some other tools retain their app plan gates, such as Agent Analytics on Scale and above.
: Same data, different interface. The REST API is for code you write and HTTP requests you make. MCP lets an AI agent reach the same metrics and actions conversationally, choosing the right tool on its own. Use whichever fits your workflow, or both.
: Yes. LLM Pulse provides a hosted MCP server that works with ChatGPT, Claude, Gemini, Cursor and any MCP-compatible client. It includes about 90 tools for reading data and taking action, connects through secure OAuth with no API key, and is available on every plan.

More ways to integrate LLM Pulse

API Access

The same data and actions exposed as a documented REST API.

Looker Studio Connector

Visualize AI visibility metrics with our ready-made Looker Studio template.

Web Analytics Integration

Blend AI visibility with your GA4, Plausible or PostHog data.

Chrome Extension

Detect the AI queries your audience asks, right in the browser.

API documentation

Endpoints, auth and the full MCP tool catalog in one place.

From the blog

Solutions

Agency solutions

How agencies wire LLM Pulse into AI agents and reporting workflows.

Developers

API and MCP reference

Full documentation for the REST API and Model Context Protocol server.

Pricing

See pricing and plans

API and MCP access across the LLM Pulse self-serve and Enterprise plans.

Ready to connect your AI assistant?

MCP is included free on every plan. Start a 14-day trial and connect your AI assistant in minutes.