AI status in standup
You ask
“Check my brand visibility across ChatGPT and Perplexity”
Agent returns
Table: Appearance, Cited / Mentioned / Absent, gaps, SoV — with asOf and confidence.
UX Report
Integrations · MCP
Cursor, Claude, and ChatGPT pull live AI status, propose the fix, and re-check the answers — on the same server as the product.
How it works
The assistant asks in plain language. UX Report answers with monitored truth, then the path to apply.
01
The assistant your team already has open sends the request.
02
One HTTPS server maps the ask to the Agent API. Same tools in every client.
03
Returns status with asOf, the Opportunity, placement, and a re-check.
> user: What's my AI status?
> mcp: invoking get_ai_status
> platform: { asOf: "…", appearance: 42%, cited: 18%, gaps: 11 }
> assistant: Here's the table with asOf — 11 buyer questions still absent.
Setup
Cursor and Claude Code take a key. Claude and ChatGPT send you back to approve a project.
In the app: Settings → Integrations. Pick Cursor, Claude, or ChatGPT.
Open Settings → Integrations →Paste the config for Cursor or Claude Code. For Claude and ChatGPT, approve the project when UX Report opens.
Try: “What's my AI status?” — you should get a table with asOf, not a made-up score.
{
"mcpServers": {
"uxreport": {
"url": "https://app.ux.report/api/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}For developers
claude mcp add --transport http --scope user uxreport https://app.ux.report/api/mcp --header "Authorization: Bearer YOUR_API_KEY"claude mcp listWhat's my AI status?Custom connector
https://app.ux.report/api/mcpDeveloper mode
https://app.ux.report/api/mcpIn the agent
The same jobs in Cursor, Claude, and ChatGPT — mapped to real tools.
You ask
“Check my brand visibility across ChatGPT and Perplexity”
Agent returns
Table: Appearance, Cited / Mentioned / Absent, gaps, SoV — with asOf and confidence.
You ask
“How does our AI visibility compare to tracked competitors?”
Agent returns
Side-by-side SoV from the latest pass — your configured competitors only.
You ask
“Where are answer engines citing us on the latest pass?”
Agent returns
Citation hosts: owned?, prompt count, sample URL — same substrate as in-app.
You ask
“What should we fix first for the biggest coverage gaps?”
Agent returns
Prioritized proposal cards: buyer question, cause, destination pages, structural change.
You ask
“Yes — do proposals 1 and 2”
Agent returns
Accept → place from Page Model → apply payload → build in your SoR → re-check.
You ask
“Run a technical readiness check and show top issues”
Agent returns
Technical health stays in the product today. The agent tools cover status, proposals, placement, apply, and re-check.
Agent tools
Monitor → decide → place → apply → measure. Server name uxreport, in every assistant.
get_ai_status
LiveAI Visibility table + asOf
propose_solutions
LiveIntent Diagnosis proposal cards
accept_solutions
LiveCreate Opportunities on consent
propose_placement
LiveAuto section from Page Model
approve_placement
LiveApprove where it lands
get_apply_payload
LiveWHERE + WHAT for the build
report_apply_status
LiveWrite-back from the agent
trigger_recheck
LiveMeasure after apply
technical_readiness
SoonCrawl / read readiness for AI
Cursor
Remote MCP · API key
Claude Code
Remote MCP · API key
Claude
OAuth connector
ChatGPT
OAuth · Developer mode
Security
Cursor and Claude Code hold a key. Claude and ChatGPT approve access in UX Report.
Technical details
Cursor, Claude Code, Claude, and ChatGPT. They call the same server and the same eight tools.
Settings → Integrations. Cursor and Claude Code need a key. Claude and ChatGPT approve a project.
No. In ChatGPT turn on Developer mode under Settings → Security and login, then add a plugin from Browse plugins with the server URL and OAuth.
Developer mode is on paid ChatGPT plans on the web. Cursor and Claude Code only need your UX Report key.
Yes. One project per client. Lock the key, or the OAuth approval, to that project.
Agent tools for that project: AI status, proposals, Opportunities, placement, apply payload, apply status, and re-check.
Settings → Integrations → the assistant you connected → revoke the key. The next call fails.
No. The product stays the record for monitoring, Opportunities, and measurement. Assistants run the same loop.
Connect Cursor, Claude, or ChatGPT. Ask for AI status. Approve before anything ships.