Run naia from
the agent you
already use.
naia speaks MCP. Connect Claude Code, Cursor, Codex, Windsurf, VS Code, Replit or Lovable to today.naia/platform and ship GEO analysis, execution plans and humanized content without leaving the conversation.
What a call looks like
The same JSON-RPC every client speaks.
One server, one protocol. Ask the assistant to list tools, prepare a brand setup, then kick off an analysis. The server returns IDs the assistant can poll.
Request
{
"jsonrpc": "2.0",
"id": 1,
"method": "tools/call",
"params": {
"name": "autopilot_start_analysis",
"arguments": {
"brandId": "abc123",
"models": ["gemini", "chatgpt"]
}
}
}Response
{
"jsonrpc": "2.0",
"id": 1,
"result": {
"runId": "run_xyz",
"status": "pending",
"creditsReserved": 8,
"pollWith": "autopilot_get_run_status"
}
}Six rhythms, one server
Who runs naia from their editor.
Marketing teams
Brief the assistant, watch it draft a quarter of humanized GEO content with the plan attached.
autopilot_create_execution_planNewsroom and editorial
Schedule analyses across ChatGPT, Gemini, Claude and Perplexity. Read the diff in your editor.
autopilot_start_analysisE-commerce
Audit category coverage in AI search, generate landings where citations drop.
autopilot_supermemory_scanAgency and studio
Switch brand setups, pause an agent for one client, queue a manual run for another.
autopilot_list_brandsBrand custodian
Persist tone, banned phrases and product positioning. The assistant pulls them before every run.
autopilot_memory_createDevtools and SaaS
Call the same MCP tools from your backend through JSON-RPC. Ship AI visibility inside your app.
JSON-RPC 2.0The 19 tools
Every tool the assistant can call.
autopilot_list_brandsList Autopilot brands available to the API key
freeautopilot_prepare_brand_setupGenerate the brand setup and approval document before any run
freeautopilot_start_analysisStart an approved Enterprise GEO analysis
creditsautopilot_get_run_statusPoll an Autopilot run until it completes
freeautopilot_get_reportRead report, visibility and technical audit data
freeautopilot_export_reportExport a finished report as CSV, Excel or PDF
freeautopilot_create_execution_planGenerate a GEO execution plan with phases and tasks
10 creditsautopilot_get_execution_planRead plan, phases and task progress
freeautopilot_generate_contentGenerate humanized GEO content with brand voice
50 creditsautopilot_get_contentRead generated content status and draft
freeautopilot_update_contentEdit saved content draft, briefing, title, status or collection
freeautopilot_run_technical_auditRun technical GEO audit on the brand site
freeautopilot_memory_listList brand memories used by every generation
freeautopilot_memory_createCreate or update a brand memory entry
freeautopilot_supermemory_scanCreate a SuperMemoria scan over the brand domain
freeautopilot_supermemory_searchSearch SuperMemoria nodes by topic
freeautopilot_get_creditsRead credits and per-operation cost estimates
free
Install in 30 seconds
Pick your client, paste the snippet.
Remote HTTP for clients that support it, npm stdio for command-based clients. Clients with MCP OAuth support (Claude Code, claude.ai, Cursor, VS Code, ChatGPT) connect with universal login: add the URL with no headers and approve in the browser, no API key to copy. Sessions last up to 180 days and can be revoked anytime in the dashboard. The same server is also listed in the official MCP Registry as today.naia/platform.
# Universal login (OAuth): your browser opens to sign in and approve
claude mcp add --transport http naia \
https://naia.today/api/v1/mcp
# Alternative with API key (headless automations)
claude mcp add --transport http naia \
https://naia.today/api/v1/mcp \
--header "Authorization: Bearer naia_YOUR_KEY"How it works
Four moves, from key to first run.
Connect with your account
Add the server URL and your client opens the browser to sign in and approve (OAuth). Prefer headless automation? Generate an API key in /dashboard/settings/api.
Connect the client
Paste the snippet from the install tab. The assistant picks up 19 tools automatically.
Prepare and approve
Ask the assistant to prepare the brand setup. Approve the document. No surprise charges.
Run and ship
Trigger analyses, draft execution plans, generate humanized content. Polling stays in chat.
Also available
REST API for backend integrations.
The MCP server is the primary surface. The same operations are also available as REST endpoints when you embed naia in your own product.
POST/api/v1/analysesPOST/api/v1/contentPOST/api/v1/execution-plansGET/api/v1/businessesGET/api/v1/creditsPOST/api/v1/mcp
FAQ
Four questions before you connect.
The assistant calls naia tools from the conversation, with structured arguments and consistent brand memory. You stay in the editor, you keep the audit trail, the assistant does the orchestration.
No. Read tools execute. Paid tools require an approval document first (autopilot_prepare_brand_setup), then explicit go-ahead. Costs are shown before confirmation.
Any HTTP MCP client. Tested with Claude Code, Cursor, Codex, Windsurf, VS Code, Replit and Lovable. The npm stdio proxy covers command-based clients.
Not anymore. Clients with MCP OAuth support connect with universal login: the server replies 401, the client discovers the OAuth endpoints, opens the browser, you sign in and approve. The session lasts up to 180 days via refresh token and can be revoked anytime in the dashboard. API keys remain available for headless automations.
Yes. The same operations are available as REST endpoints under /api/v1, registered in the OpenAPI specification. Use REST when you embed naia in your own backend.