The Agent Skills Directory

[SAFE]: The skill utilizes official Google Cloud libraries (cxas_scrapi, google-genai) to interact with Conversational AI Service (CXAS) resources. All operations are performed using the user's local gcloud authentication context.
[SAFE]: Data ingestion processes (fetching evaluations, tool schemas, and agent configurations) target authenticated cloud resources within the user's project environment.
[SAFE]: Script execution involves standard data transformation and report generation. The implementation of run_evals.py includes proper HTML escaping for generated reports and uses ast.literal_eval for safe data parsing.
[SAFE]: The skill does not contain any obfuscated code, hardcoded credentials, or unauthorized network exfiltration patterns. It follows the principle of least privilege by relying on existing environment authentication.

cxas-sim-eval