Lastfind runs an MCP server, so Claude, ChatGPT, Cursor and other AI apps can read your reports and, if you allow it, make safe changes for you. Add this address to the app:
https://lastfind.example.com/mcpThe app opens Lastfind in your browser: sign in, review the request and approve. Reading is always included; changes only if you tick Allow changes. The connection then appears in Settings > AI apps.
Claude#
On claude.ai or the desktop app, open Customize > Connectors, choose + Add and then Add custom connector. Name it Lastfind, paste the address, continue and keep the suggested sign-in. On Team and Enterprise plans an owner adds it under Organization settings > Connectors, and each person clicks Connect.
In Claude Code, run this and then /mcp to sign in:
claude mcp add --transport http lastfind https://lastfind.example.com/mcpChatGPT#
Custom MCP servers need developer mode, which your plan or workspace admin may have to allow (Settings > Apps > Advanced settings). Then open chatgpt.com/plugins, choose + and Add custom MCP server, paste the address, choose OAuth and create it. ChatGPT renames these menus often; if yours differ, look for custom connectors or MCP servers.
Cursor#
Add Lastfind to .cursor/mcp.json in a project, or to ~/.cursor/mcp.json for every project. Cursor opens the browser to sign in the first time.
{
"mcpServers": {
"lastfind": {
"url": "https://lastfind.example.com/mcp"
}
}
}Personal access tokens#
For an app or script that cannot sign in on its own, create a token in Settings > AI apps > New token: name, workspace, permission and expiry (30 days, 90 days or 1 year), up to 20 tokens at a time. It is shown only once. The app sends it in the Authorization header, for example in Cursor:
{
"mcpServers": {
"lastfind": {
"url": "https://lastfind.example.com/mcp",
"headers": {
"Authorization": "Bearer ${env:LASTFIND_TOKEN}"
}
}
}
}Keep the token in the LASTFIND_TOKEN environment variable, never in a shared file.
What apps can do#
Read, with any connection: projects, overview, prompts, answers, brand comparison, discovered brands, sentiment, sources and the detail of each source, source types, AI models, query fan-out, products, opportunities and Semantic Branding with its evidence, with the same filters and periods as the app. Missing data comes back as missing, never as zero.
Change, only with Allow changes and a member, admin or owner role: create, edit, pause, resume, archive and restore prompts; organize them with topics and tags; create, rename and delete empty topics; confirm, reject, rename, merge or undo discovered brands. Apps ask you before each change, and every change is listed in Settings > AI apps > Recent actions.
Never: plans, billing, members, project settings, creating or archiving projects, backups, exports, refreshing analyses or administration.
Only creating or resuming prompts starts monitoring: they are collected exactly like prompts added in the app (in the next scheduled run on Cloud, right away in an independent installation), within the same limits. Reading never calls an AI or a data provider.
Disconnect and revoke#
Settings > AI apps lists every connected app and token with its workspace, permission and last use. Disconnect or Revoke cuts access at once. Tokens expire on their date, app connections after 30 days without use, and a person removed from a workspace loses access through every app.