The Agent Skills Directory

[DATA_EXFILTRATION]: The get command in scripts/env_manager.py retrieves raw, unmasked environment variables from both the specified file and the system's os.environ. This enables the extraction of sensitive secrets like API keys or database URLs directly into the agent's output.
[DATA_EXFILTRATION]: The scripts/env_manager.py script allows reading from and writing to arbitrary file paths provided via the --file argument. This can be abused to access sensitive configuration files beyond the intended .env files if they follow a compatible format (e.g., ~/.aws/credentials).
[COMMAND_EXECUTION]: The skill requires the execution of a local Python script (scripts/env_manager.py) which performs sensitive file system operations and accesses the system environment.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external .env files without sanitization or boundary markers, creating a surface for indirect prompt injection if the agent interprets the loaded values as instructions.
Ingestion points: parse_env_file in scripts/env_manager.py reads content from user-specified files.
Boundary markers: Absent. The script parses all line content as key-value pairs.
Capability inventory: File system write (write_env_file), system environment access (os.environ), and local script execution.
Sanitization: Absent. Values are read and output directly without filtering or escaping.

env-manager