The Agent Skills Directory

Command Execution for Diagnostics: The skill performs read-only operations using the gcloud CLI and curl to gather evidence from the user's project. It adheres to a strictly defined workflow and requires explicit user consent before executing any state-changing commands.
Trusted External Dependencies: For OpenAPI validation, the skill utilizes a local Python script that may depend on the PyYAML package. This is a standard library from a well-known source, used here to facilitate local parsing of configuration files.
Data Ingestion Considerations: The skill analyzes user-provided OpenAPI documents and project log entries. This represents an indirect prompt injection surface; however, the provided linter script employs yaml.safe_load to ensure safe data handling. The agent is also instructed to redact sensitive information before displaying results.
Telemetry and Attribution: Following vendor practices, the skill includes headers and environment variables in its requests to allow for usage attribution. This is a transparent mechanism for monitoring skill performance within the Google Cloud ecosystem.

api-gateway-troubleshooting