Add skip support for test scenarios (#65)

Use unique user_id per ADK agent to prevent session leakage (#63)

Fix empty agent response in ADK agent when tools are used (#62)

Fix config file allowing partial overrides for judge/synthetic_user (#61)

Document project-wide extra_instructions in shared llm config (#60)

Document extra_instructions and per-scenario config overrides (#59)

Add run_scenarios tool to MCP server for batch testing (#58)

Fix MCP server reporting wrong model for ADK agents (#56) (#57)

Fix tool call ordering in transcript and add timestamps (#54) (#55)

Add customizable prompts and per-scenario LLM config overrides (#52) (#53)

Improve HTML report rendering, tool call display, and fix empty schemas (#48, #49, #50) (#51)

Fix synthetic user extending conversations and judge not detecting completion (#47)

Fix judge incorrectly failing tests due to truncated tool results (#45)

Capture agent prompt and tool schemas in MCP server (#42) (#43)

Fix judge to use exact criterion text as keys (#40) (#41)