CATT - Surrogate agent testing for MCP servers and APIs in CI
Show HN: CATT 0.2.1: Test agents/services against a Muse/Instinct-alike, in CI
CATT (Consumer Agent Testing Tool) lets you test how an AI assistant actually uses your service. Run scripted or model-driven agents against MCP servers, MCP apps, websites, or HTTP APIs, then verify what the service did, inspect sealed evidence, and diff results across builds. It emulates Muse and Instinct hosts through a shared connector testing contract, covering routing, memory, payments, browser tasks, and scheduling. Scripted runs need no model account, and quickstarts run on loopback with no configuration. Ideal for catching integration regressions before they reach real assistants.
CATT tests how an assistant uses your service: run an agent, check what the service did, inspect the evidence, and compare results across builds.