The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data such as SQL DDL files and JSON schema definitions. This creates an attack surface for indirect prompt injection, where malicious instructions could be embedded in schema metadata or identifiers to influence the agent's interpretation of the results.
Ingestion points: The tools schema_analyzer.py, index_optimizer.py, and migration_generator.py read external file content through command-line arguments.
Boundary markers: The scripts do not currently implement specific delimiters or instructions to prevent the agent from following commands potentially embedded within analyzed SQL comments or data types.
Capability inventory: The skill is restricted to local file processing and text/SQL generation. It does not include network access, arbitrary shell execution, or sensitive file system write capabilities.
Sanitization: Input content is parsed using regular expressions and standard JSON loaders, but reflected content in the output reports is not sanitized for potential natural language instructions.

database-designer