Integrations
MCP server
Connect Claude, or any MCP client, straight to your GenAuthority workspace.
GenAuthority ships a Model Context Protocol server, so an AI assistant can read your visibility data and make changes to your workspace without you copying anything into a chat window.
Endpoint
POST https://api.genauthority.com/mcpJSON-RPC over Streamable HTTP. Authorisation is OAuth: your client sends you through a consent screen, you approve, and it receives a token scoped to your workspace. Tokens can be revoked at any time from Team → API access.
The tools
Thirteen, split by whether they change anything.
Read (9)
| Tool | Returns |
|---|---|
list_brands | Every brand you can see. Call this first, the rest take a brand from it. |
get_visibility | Visibility percentage, overall and per engine |
get_share_of_voice | Your share against tracked competitors |
get_prompts | The tracked prompt library, with paused state |
get_perception | Attributes engines associate with the brand |
get_citations | Cited sources from the latest scan |
get_outreach_targets | The citation gap, as a ranked list |
get_paid_placements | Ads observed inside AI answers |
get_fanout_queries | The searches an engine ran before answering |
Every read returns figures from the brand’s most recent completed scan, and each response states when that was. A number without a date is a number you cannot act on.
Write (4)
| Tool | Does |
|---|---|
create_brand | Adds a brand |
update_brand | Edits name, website or competitors |
add_prompt | Adds a tracked prompt |
add_competitor | Names a competitor |
The write tools create rows and nothing else. They never trigger an engine call, never spend your run budget, and everything they do is visible and undoable in the app. That boundary is deliberate: an assistant should be able to set your workspace up, not to spend your money.
Permissions
The server reads your role live rather than trusting what was true when the token was minted. If your access is reduced, the next call reflects it. A read-only client role gets the read tools and is refused the write ones.
Setting it up in Claude
Add the endpoint as a remote MCP server in your client’s connector settings and complete the OAuth flow when prompted. Once connected, ask it something like:
Which sources cite my competitors and not me?
and it will call list_brands, then get_outreach_targets, and answer from
your actual scan data.
REST
Everything the MCP server reads is also available over the REST API with a workspace API token, for scripts and dashboards that are not conversational. Create one under Team → API access.