MCP + HTTP API access is included on Free. Connect your AI in about two minutes →
Developer documentation

Make search intelligence callable.

Create a token, connect through MCP or the HTTP API, and let the model retrieve current SearchSignal workspace data.

01

Quickstart

Create a SearchSignal account, add a project, run or import the data you need, and issue a token from the authenticated Connect AI area. Use the generated client config rather than copying a token from screenshots or logs.

The public examples use YOUR_TOKEN. Never place a real token in public HTML, analytics events, screenshots, or a repository.
  1. Create or sign in to a SearchSignal workspace.
  2. Add the domain and optionally connect Google Search Console.
  3. Run an audit or save the keyword and competitor research the agent will need.
  4. Create an MCP/API token.
  5. Paste the exact config for the selected client.
  6. Ask for latest_audit, list_opportunities, explore_keyword, or workspace_pack.
02

Authentication

MCP clients receive the credential through the client configuration or environment. HTTP clients send the token using Bearer authorization.

HTTP authorization header
Authorization: Bearer YOUR_TOKEN

Token lifecycle

  • Create: issue the first token from the Connect AI screen.
  • Rotate: invalidate the old credential and reveal a new one.
  • Revoke: remove access when a client or environment should stop using the workspace.
03

MCP setup

For clients that support Model Context Protocol, SearchSignal uses an npx-compatible package path and the workspace token.

Generic MCP configuration
{
  "mcpServers": {
    "searchsignal": {
      "command": "npx",
      "args": [
        "-y",
        "https://searchsignal.pro/mcp/searchsignal-mcp-1.0.1.tgz"
      ],
      "env": {
        "SEARCHSIGNAL_TOKEN": "YOUR_TOKEN"
      }
    }
  }
}
Client schemas and setup locations can differ. The authenticated Connect AI page should generate the exact configuration for Cursor, Claude Desktop, Claude Code, Windsurf, VS Code, Continue.dev, Cline, and Zed.
04

HTTP API

Use the HTTP path for ChatGPT workflows, custom models, automations, and systems where MCP is not the chosen transport.

Inspect available tools
curl https://searchsignal.pro/api/mcp/v1/tools \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json"

Call a tool

Tool call example
curl -X POST https://searchsignal.pro/api/mcp/v1/call \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "tool": "latest_audit",
    "arguments": {
      "project": "example.com"
    }
  }'

The final production docs should be generated from the live tool contract so endpoint paths, parameters, response shapes, and errors cannot drift from the backend.

05

Core tools

latest_audit

Return the most recent crawl summary and scored findings for a selected project.

list_opportunities

Return prioritized, filterable opportunities with status, impact, effort, and supporting evidence.

explore_keyword

Return keyword metrics, intent, SERP context, related ideas, and opportunity scoring.

workspace_pack

Return a compact agent-ready briefing that combines the most useful project context for a task.

SearchSignal's product surface also includes keyword ideas, bulk analysis, domain exploration, competitor gaps, rank tracking, backlinks, content briefs, and AI writing. The live /tools response should remain the authoritative catalog of callable tool names.

06

Prompt patterns

Audit to implementation

Pull the latest audit for example.com.
Rank the open opportunities by expected impact divided by effort.
Then inspect this repository and create a file-by-file implementation plan.

Keyword to page strategy

Explore the keyword “SEO MCP server.”
Return intent, opportunity, SERP pattern, supporting terms,
and the page type most likely to satisfy the query.

Workspace briefing

Give me a workspace_pack for example.com.
Summarize current health, strongest traffic opportunities,
ranking momentum, and the next five actions.
07

Security

  • Never commit tokens to source control.
  • Do not expose full token values in browser analytics, server logs, screenshots, or support tickets.
  • Rotate credentials after suspected exposure or when a shared environment changes ownership.
  • Keep token issuance and storage server-side in the production app.
  • Rate-limit and audit calls at the authenticated workspace boundary.
  • Return only the project data the signed-in workspace is authorized to access.
08

Production documentation checklist

  • Generate the tool catalog from the live MCP/API schema.
  • Document request and response examples from real test fixtures.
  • Publish stable error codes and retry guidance.
  • Version the package and API contract deliberately.
  • Add a changelog when tool names, arguments, or response shapes change.
  • Keep client setup screenshots current as Cursor, Claude, Windsurf, and VS Code interfaces evolve.
Connect once. Ask anywhere.

Connect your first client.

Create the token, use the generated client config, and ask for a real project briefing from Cursor, Claude, ChatGPT, or your own agent.