The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by capturing and later displaying potentially untrusted data.
Ingestion Points: Data from external security tools enters via StorageAPI methods in scripts/storage_api.py from tools like nmap or sqlmap.
Boundary Markers: Reports generated by ReportGenerator do not include delimiters or warnings to isolate untrusted data from the agent's instructions.
Capability Inventory: The skill has filesystem write access to manage its SQLite database.
Sanitization: The skill uses parameterized SQL for database integrity, but lacks natural language instruction filtering or escaping in its reporting output.
[SAFE]: The skill implements robust security measures for local data storage.
Evidence: scripts/db_manager.py enforces restrictive file system permissions (0o700 for the data directory and 0o600 for the database file).
Evidence: All database operations in scripts/storage_api.py utilize parameterized queries to prevent SQL injection vulnerabilities.

results-storage