The Agent Skills Directory

[COMMAND_EXECUTION]: The Python utility scripts/generate_config.py is capable of generating shell scripts from arbitrary content and setting executable permissions using chmod 0o755. This creates a pathway for the dynamic generation and execution of code within the agent's environment.\n- [DATA_EXFILTRATION]: Instructions in SKILL.md require the agent to search for and read sensitive files, including .env files, Kubernetes Secrets, and source code containing hardcoded credentials. This access is inherent to configuration management but exposes sensitive credentials to the agent's context.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection through its ingestion of untrusted local data.\n
Ingestion points: Audit step in SKILL.md reading .env files, configuration directories, and source code.\n
Boundary markers: None present to distinguish data from instructions.\n
Capability inventory: File writing and execution capabilities via scripts/generate_config.py and the Bash tool.\n
Sanitization: No escaping or validation is performed on the ingested content before processing.

managing-environment-configurations