The Agent Skills Directory

Indirect Prompt Injection Surface: The skill facilitates querying BigQuery tables and ingesting the results into the agent's context. This creates a potential surface where data stored in the database could contain instructions designed to influence the agent's actions. This is a common consideration for skills that process external data.
Ingestion points: SQL query results from commands like bq query and metadata retrieved via MCP tools such as get_table_info.
Boundary markers: The instructions do not define specific delimiters or warnings for the agent to use when handling data results.
Capability inventory: The skill utilizes shell command execution for BigQuery management (bq), infrastructure configuration with Terraform, and package installation.
Sanitization: There are no explicit steps provided for sanitizing or validating the data returned from BigQuery before it is processed by the agent.
Official Library Dependencies: The documentation includes instructions for installing official Google Cloud client libraries, such as google-cloud-bigquery and @google-cloud/bigquery. These are standard, vendor-provided resources for interacting with the service.
Usage Attribution Telemetry: The skill instructs the agent to include attribution tags in CLI environment variables and HTTP User-Agent headers. This is a standard mechanism for cloud providers to track the usage of specific integration tools and does not involve the collection of sensitive user credentials or data.

bigquery-basics