Standardized YAML Format: This will provide a static way to express your Trailblaze tests mixing natural language
objectives and static steps
Recording: Save any agent tools and ui interactions, then play them back to save costs during future runs.
Test Failure Recovery with AI: When a recorded test fails, use the AI Agent to recover and complete the objectives.
Then leverage the reporting tools to show divergence from the original path.
MCP Server: Interactively create new tests and manage your test suite.
Bring your own LLM Provider: OpenAI support currently, but future plans for other providers.
LLM call proxying support (optional): Provide support for proxying traffic to the LLM, removing the requirement of
passing an API key the mobile device as an execution argument.