The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project source files through its reconnaissance and risk-scanning scripts, creating a surface for potential instruction injection.
Ingestion points: The scripts scripts/inspect_database_project.py and scripts/scan_schema_risks.py read and analyze all source files within the project root.
Boundary markers: The scripts output structured JSON data to the agent, providing some structural separation, though no explicit 'ignore embedded instructions' delimiters are used when processing file content.
Capability inventory: The skill enables the agent to read file contents, perform static analysis, and suggest or implement database schema and migration changes.
Sanitization: The scripts use regular expressions to extract metadata and detect risks but do not perform sanitization or filtering of the raw text before it is presented to the agent.

database-engineering