$ man mcp-litmus-test

MCP Litmus Test

A three-level evaluation framework for GTM tools: Level 1 - does it have an API? Level 2 - is there a CLI? Level 3 - does it ship an MCP server? The score determines how automatable the tool is and whether it belongs in a modern GTM stack.

every GTM tool has a web interface. that is table stakes. the real question is whether the tool can be operated without clicking. can an AI agent call it? can a cron job trigger it? can a script run it at 2 AM? a tool stuck at GUI-only has an automation ceiling. you can scale the team, but you cannot scale the process. I evaluate every tool in a client's stack against this test. the results usually reveal why certain workflows bottleneck - the enrichment layer is fully automatable (Level 3) but the outreach layer requires manual intervention (Level 0). that mismatch is where pipeline velocity dies.

I score every tool in the stack 0-3 during a stack audit. Level 0 means GUI only - no programmatic access at all. Level 1 means REST API with authentication. Level 2 means official CLI tooling. Level 3 means MCP server available. tools like HubSpot and GitHub score 3. tools like Clay score 1-2 (API exists but most power is in the GUI). tools with no API score 0 and get flagged for replacement. the aggregate score across the stack tells you how automatable your GTM motion is. a stack averaging 2.5+ is ready for agent-driven orchestration. a stack averaging 1.0 needs infrastructure work before automation makes sense.