Loading...
mutiro agent evals run

Send each eval case to the agent and judge the replies

Usage mutiro agent evals run <evals-dir> [flags]
Options
Flag Type Description
--agent string Agent username to evaluate (required)
-h, --help bool help for run
--in-conversation-with string Probe the agent's conversation with this user (owner interjection): the agent answers as that conversation's self
--only string Run a single case by name
--role string Run cases written for this role: owner or user (match your login, or use --in-conversation-with) (default "owner")
Global Options
Flag Type Description
--config string config file (default is ./.mutiro.yaml in current directory)
--debug bool debug mode
-v, --verbose bool verbose output
Subcommands
See Also
mutiro agent evals