About GetIntel
GetIntel is an AI visibility platform connector for brands using MCP clients. It exposes brand performance data and lets agents review opportunities and approve actions, with access scoped by account and brand.
Key features
- 01AI visibility scores and trends
- 02Buyer-question and engine coverage
- 03Competitor and citation-source analysis
- 04Content gaps and crawlability checks
- 05Review and approve missions or findings
Use cases
- 01Track how AI engines mention a brand
- 02Find missing content for buyer questions
- 03Compare brand visibility with competitors
- 04Approve suggested optimization work
MCP support
A remote MCP server at `https://app.getintel.ai/mcp` exposes read and action tools to MCP clients. OAuth 2.1 controls scopes, and access is limited to the authenticated account and brand.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://app.getintel.ai/mcpclaude mcp add --transport http getintel https://app.getintel.ai/mcp{"mcpServers":{"getintel":{"type":"http","url":"https://app.getintel.ai/mcp"}}}Open in Cursor{"servers":{"getintel":{"type":"http","url":"https://app.getintel.ai/mcp"}}}Open in VS Code{"mcpServers":{"getintel":{"type":"http","url":"https://app.getintel.ai/mcp"}}}{"mcpServers":{"getintel":{"serverUrl":"https://app.getintel.ai/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEGetIntel MCP Connector
GetIntel is an AI visibility (GEO) platform — it tracks how often ChatGPT, Perplexity, Gemini, and Google AI Overviews name your brand in buyer conversations, finds the gaps, and drafts the fixes. This connector lets your own AI agent (Claude, or any MCP client) read that data and act on it directly, in conversation.
- Server:
https://app.getintel.ai/mcp(remote,streamable-http) - Auth: OAuth 2.1 (Dynamic Client Registration + PKCE) — no pasted API keys
- Registry:
io.github.tarang8811/getintelon the Official MCP Registry - Privacy policy: https://app.getintel.ai/privacy
- Requires: an active GetIntel account
This repository holds the connector's public documentation and tool manifest. The server implementation lives in GetIntel's main application repo.
Connecting
- In your MCP client, add a remote server with URL
https://app.getintel.ai/mcp. - Complete the OAuth flow — you'll be redirected to sign in / authorize with your GetIntel
account and choose which scopes to grant (
read,act). - The client discovers available tools automatically via
tools/list— a read-only token only sees thereadtools below.
Each brand in your account gets its own token, scoped so an agent connected for one brand can
never see another brand's data — see tools.json for the full schema.
Tools
Every call is scoped to the authenticated user/brand. read tools never mutate anything.
get_visibility- AI Visibility score (0-100) — trailing 7-day citation rate, with trend and history.get_ai_answers- Buyer questions × engine grid — where the brand is open, absent, mentioned, or recommended.get_receipts- The exact AI answer for one buyer question, across engines.list_activity- Recent things the agent found or did, most recent first.list_missions- Proposed/active batches of related opportunities awaiting approval.get_mission- Read-only preview of one mission before approving it.get_competitors- Share-of-voice vs named competitors over 30 days.get_prompt_matrix- Per-prompt × per-engine state matrix.get_sources- Domains AI engines cite in your category — where to earn citations.get_content_gaps- Per-question coverage: do you have a page, does AI cite it, what to do.get_signals- Reddit/X threads AI cites where you're absent, or competitor complaints.get_search_console- Google Search Console performance joined against tracked AI prompts.get_authority- Domain rating, referring domains, and link-building opportunities.get_brand- Does AI trust/recommend the brand by name — the reputation signal.get_brand_transcript- Full untruncated prompt/answer/citation text behindget_brand.get_crawlability- Can AI crawlers actually reach and read the site (12 pass/warn/fail checks).get_usage- This month's MCP read/draft budget and when it resets.get_prompt_history- Buyer questions won/lost vs the prior 7-day window.list_drafts- Drafts the agent has written (outreach, articles, llms.txt, replies).get_draft- Full content of one draft.approve_mission- Approve and execute a mission's whole batch. Not reversible once run.dismiss_mission- Dismiss a proposed mission without running it.replan- Re-scan current signals and propose fresh missions now.approve_event- Approve one individual finding, writing its draft.dismiss_event- Dismiss one individual finding.
Read
| Tool | Description |
|---|---|
get_visibility | AI Visibility score (0-100) — trailing 7-day citation rate, with trend and history. |
get_ai_answers | Buyer questions × engine grid — where the brand is open, absent, mentioned, or recommended. |
get_receipts | The exact AI answer for one buyer question, across engines. |
list_activity | Recent things the agent found or did, most recent first. |
list_missions | Proposed/active batches of related opportunities awaiting approval. |
get_mission | Read-only preview of one mission before approving it. |
get_competitors | Share-of-voice vs named competitors over 30 days. |
get_prompt_matrix | Per-prompt × per-engine state matrix. |
get_sources | Domains AI engines cite in your category — where to earn citations. |
get_content_gaps | Per-question coverage: do you have a page, does AI cite it, what to do. |
get_signals | Reddit/X threads AI cites where you're absent, or competitor complaints. |
get_search_console | Google Search Console performance joined against tracked AI prompts. |
get_authority | Domain rating, referring domains, and link-building opportunities. |
get_brand | Does AI trust/recommend the brand by name — the reputation signal. |
get_brand_transcript | Full untruncated prompt/answer/citation text behind get_brand. |
get_crawlability | Can AI crawlers actually reach and read the site (12 pass/warn/fail checks). |
get_usage | This month's MCP read/draft budget and when it resets. |
get_prompt_history | Buyer questions won/lost vs the prior 7-day window. |
list_drafts | Drafts the agent has written (outreach, articles, llms.txt, replies). |
get_draft | Full content of one draft. |
Act
| Tool | Description |
|---|---|
approve_mission | Approve and execute a mission's whole batch. Not reversible once run. |
dismiss_mission | Dismiss a proposed mission without running it. |
replan | Re-scan current signals and propose fresh missions now. |
approve_event | Approve one individual finding, writing its draft. |
dismiss_event | Dismiss one individual finding. |
act tools that send outbound email or publish content still require the account's explicit
send/publish mandate — a connected agent can never grant itself that autonomy.
Support
Issues or questions: [email protected]
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.