ClickClacks MCP

Connect an AI agent to read-only, token-scoped product analytics with provenance on every result.

Endpoint and authentication

POST https://app.clickclacks.io/api/v1/mcp
Authorization: Bearer cka_live_...

The server uses stateless Streamable HTTP. Create an MCP & analytics key under Settings → API keys; Settings → MCP has copy-ready setup for your endpoint. Never put the raw key in source control, prompts, logs, or screenshots.

Claude Code

claude mcp add --transport http clickclacks https://app.clickclacks.io/api/v1/mcp --header "Authorization: Bearer cka_..."

Generic client

{
  "mcpServers": {
    "clickclacks": {
      "type": "http",
      "url": "https://app.clickclacks.io/api/v1/mcp",
      "headers": { "Authorization": "Bearer cka_..." }
    }
  }
}

Tools

Tool	Reads
`overview`	KPIs and visitors series
`list_funnels`	Saved funnel definitions and available summaries
`funnel_results`	One saved funnel in its persisted execution mode
`retention`	Daily, weekly, or monthly cohort grid
`list_events`	Event rows with search and cursor pagination
`event_breakdowns`	Pages, referrers, countries, and event names with event and people counts
`list_segments`	Saved segment definitions and available counts
`list_sources`	Sources, windowed counts, and health
`friction_findings`	Observed friction findings by state

Every tool accepts paired optional start and end dates in YYYY-MM-DD form and defaults to the last 30 calendar days in the project’s reporting time zone.

Agency keys: every project at once

An agency owner or admin can create an agency key (cka_agency_...) in Agency settings → MCP keys. It is a Pro feature. It uses the same endpoint and reads only the projects its creator can open, checked again on every call: remove the creator from a project, or from the agency, and the key stops reading it. It is read-only, and every call is logged against the project it read.

claude mcp add --transport http clickclacks https://app.clickclacks.io/api/v1/mcp --header "Authorization: Bearer cka_agency_..."

Tool or argument	What it does
`list_projects`	The projects the key can read: name, slug, id, time zone, sources and events this month
`project`	Every tool above takes it: a project slug or id. Required unless the key reads exactly one project. Answers name the project and its time zone.
`compare_projects`	Visitors or events for every project, or a chosen few, over the same dates, each counted in its own time zone

Provenance and honest unavailable states

Every successful call includes the same provenance in structuredContent and text content: resolved window, applied filters, the project the key belongs to, data freshness and cache state where known, a degraded flag and reason, and a definition hash where available. Fixture or fallback report rows are never presented as customer data; the affected data is returned as null.

Ask your agent

Which funnel step loses the most people this week? Show the calculation and cite the ClickClacks provenance.
Compare weekly retention for the last eight weeks. Stop if the response is degraded.
What are the top pages and referrers this month, and which happening friction findings overlap those paths?
With an agency key: which of my projects had fewer visitors this week than last? List the projects first, then compare them.