The Agent Skills Directory

[DATA_EXFILTRATION]: The skill's analysis component is designed to locate and read the contents of sensitive configuration and environment files.
Evidence: The _find_config_files method in SKILL.md identifies files matching the **/*.env* pattern, and the _check_security_issues method reads the full text of these files to search for potential secrets using regular expressions.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted configuration data from the local filesystem, creating a surface for indirect prompt injection attacks.
Ingestion points: ConfigurationAnalyzer and RuntimeConfigValidator read various configuration formats (JSON, YAML, .env, TOML, INI) from the project directory.
Boundary markers: The instructions do not define boundary markers or delimiters to help the agent distinguish between data and potential instructions embedded within config files.
Capability inventory: The skill has capabilities for broad file system reading, encryption/decryption operations, and real-time filesystem monitoring.
Sanitization: The skill implements schema validation using the Ajv library, which enforces data structure and type constraints on the ingested configuration.

deployment-validation-config-validate