About Cited
Cited is an MCP server for auditing whether AI assistants cite a business website in buyer-facing answers. It helps businesses inspect visibility and manage recommendations and fixes through Cited’s tools.
Key features
- 01Free audits of buyer questions
- 02Paid audits and reports
- 03Citation comparisons across audits
- 04Recommendations and generated fixes
- 05Scheduled rechecks and alerts
Use cases
- 01Check whether AI answers cite a business
- 02Compare citation changes after site updates
- 03Generate and track fixes for audit findings
MCP support
Exposes Cited’s audit, account, documentation, recommendation and fix workflows as MCP tools. It supports a hosted HTTP endpoint and a locally run server.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://mcp.youcited.com/mcpclaude mcp add --transport http com-youcited-cited https://mcp.youcited.com/mcp{"mcpServers":{"com-youcited-cited":{"type":"http","url":"https://mcp.youcited.com/mcp"}}}Open in Cursor{"servers":{"com-youcited-cited":{"type":"http","url":"https://mcp.youcited.com/mcp"}}}Open in VS Code{"mcpServers":{"com-youcited-cited":{"type":"http","url":"https://mcp.youcited.com/mcp"}}}{"mcpServers":{"com-youcited-cited":{"serverUrl":"https://mcp.youcited.com/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 READMEcited-mcp
MCP server for Cited: check whether ChatGPT, Gemini, Claude and Google AI Mode cite your website when buyers ask about what you sell, and get the fixes that get you cited. 21 tools over the Model Context Protocol: the 19 Cited tools, the same ones the hosted server at mcp.youcited.com serves, plus local login and logout.
Install
pip install cited-mcp
Or run directly without installing:
uvx cited-mcp
Usage
The hosted server is at https://mcp.youcited.com/mcp (streamable HTTP; your browser opens to log in on first use). The free_audit, account_status, agent_signup and search_docs tools work without logging in.
Quick setup: with the cited-cli installed, cited agent setup configures Claude Code, Cursor and Codex for you and prints the Claude Desktop connector steps (--dry-run to preview). It merges into existing config and backs up files before changing them.
Claude Desktop (Custom Connector — recommended)
No local install required. Claude connects to Cited's hosted MCP server directly:
- Open Claude Desktop → Settings → Customize > Connectors
- Click "+" → "Add custom connector"
- Enter URL:
https://mcp.youcited.com/mcp - Click Add
Your browser will open for authentication on first use. See Anthropic's connector guide for details.
Requires: Claude Pro, Max, Team, or Enterprise plan.
Claude Desktop (Developer Config — alternative)
If Custom Connectors aren't available, use the developer config with mcp-remote:
{
"mcpServers": {
"cited": {
"command": "npx",
"args": ["mcp-remote", "https://mcp.youcited.com/mcp"]
}
}
}
Add to Settings → Developer → Edit Config, then restart Claude Desktop.
Prerequisite: Node.js must be installed for
npx.
Claude Desktop (local stdio server)
If you prefer running the server locally:
{
"mcpServers": {
"cited": {
"command": "uvx",
"args": ["cited-mcp"],
"env": {
"CITED_TOKEN": "your-jwt-token"
}
}
}
}
Get your token by running cited login and then cited auth token with the cited-cli.
Claude Code
claude mcp add --transport http --scope user cited https://mcp.youcited.com/mcp
Then run /mcp in Claude Code and choose Authenticate. For Cited slash commands, skills and the geo-analyst agent, also install the plugin with /plugin marketplace add YouCited/cited-plugins and /plugin install cited@youcited-cited-plugins.
Cursor
Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json in your project root:
{
"mcpServers": {
"cited": {
"url": "https://mcp.youcited.com/mcp"
}
}
}
Then in Cursor: open Settings (Cmd+,) → MCP, click Login next to cited, and verify it shows a green status indicator. Use Cited tools in Agent mode.
Fallback (older Cursor versions without remote-server support) — proxy through mcp-remote, which needs Node.js (node --version should print v18+):
{
"mcpServers": {
"cited": {
"command": "npx",
"args": ["mcp-remote", "https://mcp.youcited.com/mcp"]
}
}
}
Codex
codex mcp add cited --url https://mcp.youcited.com/mcp
codex mcp login cited
Or add [mcp_servers.cited] with url = "https://mcp.youcited.com/mcp" to ~/.codex/config.toml.
Standalone
cited-mcp # stdio; uses `cited login`, CITED_TOKEN or an agent key
CITED_ENV=dev cited-mcp # another environment
The tools are defined once, in the Cited API. On start the server reads the current list from GET https://api.youcited.com/agent/tools (falling back to the copy bundled with cited-core) and runs each call through POST /agent/tools/{name}, so it always behaves like the hosted server.
Updating to new Cited features
Clients cache the tool list. After a Cited release, reconnect the connector (or restart a local cited-mcp) to pick up new tools.
Tools
Start with account_status: it says who is signed in, the credit balance, the businesses and the next step.
| Tool | What it does | Login |
|---|---|---|
free_audit | Free check of how AI answers 5 buyer questions about a site; poll with report_token | no |
agent_signup | Create an account for your human without a login; the key is saved locally, the claim link and code are returned | no |
account_status | Who is signed in, balance, businesses, prices, next step | no |
search_docs | Search the Cited docs, or read one with path | no |
buy_credits | Checkout link for a credit pack or one audit, or redeem a voucher | yes |
setup_business | Add a business from its website, or update it, its buyer personas and offerings | yes |
get_business | List businesses, or one business's profile, questions, schedule and recent audits | yes |
get_agent_brief | What to focus on next for a business | yes |
draft_questions | Draft, extend or replace the buyer questions the audit asks | yes |
quote_audit | What an audit costs, the balance and the models | yes |
run_audit | Run a paid audit; returns the price first, spends only with confirm=true | yes |
get_audit | List audits, or one audit's status and report | yes |
compare_audits | Citations won and lost against an earlier audit | yes |
schedule_recheck | Re-run the audit every N days, email and webhook alerts | yes |
get_recommendations | What to change, with item keys, evidence, status and fixes | yes |
update_recommendation | Mark an item done or dismissed, or open it again | yes |
generate_fixes | Write fix files and steps for items (or all); asks before spending | yes |
get_fixes | List fixes, or one fix's files, targets and steps | yes |
update_fix | done, undo, or verify the fix on the live site | yes |
login | Browser sign-in that stores a session on this machine (local server only) | no |
logout | Forget the session or saved agent key (local server only) | no |
Example Workflow
account_status, thenfree_auditwith the website for a free first look.agent_signup(orlogin), thensetup_businesswith the website.draft_questions, review them with the user, thenquote_audit.run_audit(the user approves the price), thenget_audituntil it finishes.get_recommendations,generate_fixes,get_fixesfor the files, thenupdate_fixwithdoneandverify.
Pricing: pay-as-you-go credits
No plans or subscriptions. 1 credit = $1; account_status and quote_audit show current prices. Purchase tools only return a Stripe checkout link for a person to open: nothing is bought until account_status shows the new balance. A short balance comes back as an error with status_code 402 and credits_needed.
Agent signup and keys
agent_signup makes an unclaimed account for an AI agent. The API key is saved on this machine (the OS keychain, or ~/.cited/credentials.json) and used for every call; it is never shown to the model. Give your human the claim link and 6-digit code so they keep the account. logout forgets the key.
Environment Variables
| Variable | Description |
|---|---|
CITED_TOKEN | A Cited session token (cited login, then cited auth token). Optional. |
CITED_AGENT_API_KEY | A Cited agent key (cak_...). Optional; agent_signup saves one for you. |
CITED_ENV | prod (default), dev or local. |
Development
pip install -e ../core -e ".[dev]"
pytest
ruff check src/ && mypy src/ --ignore-missing-imports
Privacy Policy
See youcited.com/privacy.
License
Proprietary — see youcited.com for terms.
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.