Connect over MCP

Point an AI client at your KWAYZER workspace over the Model Context Protocol. It reads; it does not write.

Every tool is a read. every tool is a read. `McpTool.mutates` is the literal type `false`, so a write tool cannot be added without changing the type — which shows up in review rather than slipping in as a line.
Your workspace comes from your key. the workspace comes from your API key and is never a parameter, so a client cannot ask for somebody else's data whatever it sends.
A refusal is an answer. a tool you may not call comes back as a refusal with the reason, not as an error. "You may not read deals" is an answer.
Engagement data is explicit. a key that can read your GTM data does NOT thereby reach a consulting engagement's semantic layer. That scope is held explicitly or not at all.

The endpoint

https://www.kwayzer.com/api/mcp

Client configuration

Replace the placeholder with a key from your workspace settings. keys are created where keys are managed, not from a documentation page. This screen tells you what a key can reach; it does not hand you one.

{
  "mcpServers": {
    "kwayzer": {
      "url": "https://www.kwayzer.com/api/mcp",
      "headers": {
        "Authorization": "Bearer kwz_live_YOUR_KEY_HERE"
      }
    }
  }
}

What your key can reach

Each tool names the permission it needs. A key you create with read access reaches the product tools below; the engagement tools need a scope you grant on purpose.

Tool	Reads	Needs
list_contacts List contacts in the workspace, filtered and paginated.	contacts	contact.view
get_contact Read one contact by id.	contacts	contact.view
list_companies List companies in the workspace.	companies	contact.view
list_deals List deals, optionally by stage or owner.	deals	deal.view
get_pipeline_summary Open pipeline value and count by stage.	deals	deal.view
list_segments List the workspace's audiences and their definitions.	segments	contact.view
get_play_metrics Read a GTM play's measured metrics, each with its class of evidence.	gtm_play_metrics	reports.view
list_governed_terms The business terms governing a migration engagement: what each means, whether that meaning is authoritative, which fields implement it, and what is unsettled about it. A contested term is returned WITHOUT a definition.	engagement_semantics	migration.viewGranted explicitly or not at all — a read key does not include it.
list_migration_datasets The datasets recorded for an engagement, with field and record COUNTS and their trust badges. Structure and counts only — no field values are ever returned.	engagement_semantics	migration.viewGranted explicitly or not at all — a read key does not include it.
get_field_meaning What one column means: its recorded type and fill rate, and the governed term implementing it, if any. Every answer carries how old the underlying measurement is.	engagement_semantics	migration.viewGranted explicitly or not at all — a read key does not include it.

Scopes an MCP key may hold

Read-only by construction — the allowlist is every scope whose name ends in :read, and writes are off at the server: MCP_WRITES_ENABLED = false.

contacts:read
companies:read
deals:read
segments:read
reports:read
mcp:read
migration:read

The full REST surface is in the API reference.