run-evals

SUSPICIOUS: The core purpose is coherent for an E2E testing skill, and use of the official Daytona CLI is plausible. However, the skill has elevated risk because it sources broad secret volumes into the test environment, depends on powerful repo-local shell scripts, can publish artifacts, and asks the agent to load another skill, creating transitive trust. This looks more like a high-risk testing/infrastructure skill than outright malicious behavior.