The Agent Skills Directory

Command Execution: The provided extraction_stats_template.py script uses subprocess.run to invoke the official gcloud CLI for listing and reading archives from Google Cloud Storage. This is a functional requirement for analyzing extraction results and uses list-based arguments to mitigate shell injection risks.
Handling of Sensitive Data: The skill demonstrates high security awareness regarding credentials. It explicitly instructs the user not to share passwords in the chat and provides specific shell commands using umask 077 to create local password files with restricted permissions. It also includes mandatory cleanup steps to delete secrets from Google Cloud Secret Manager after use.
External Resource Management: The skill utilizes official Google Cloud components (bqms) and standard JDBC drivers. It provides clear guidance on installation through official channels like gcloud components install or system package managers.
Indirect Prompt Injection Surface: As a data processing tool, the statistics script ingests and reports on metadata from external databases (e.g., table and schema names). This represents a potential surface for indirect prompt injection if the source data contains adversarial strings. The risk is minimized by the use of structured reporting and the intended use case of the diagnostic tool.
Ingestion points: extraction_stats_template.py (via analyze_zip_archive reading catalog CSVs/JSONLs).
Boundary markers: The script produces a structured Markdown report intended for agent consumption.
Capability inventory: The agent executes shell commands (gcloud), writes files, and handles sensitive credentials.
Sanitization: The script performs basic stripping but does not semantically sanitize database identifiers against potential prompt injection strings.

bigquery-migration-metadata-extractor