The Agent Skills Directory

[DATA_EXFILTRATION] (MEDIUM): The skill uses cat .env to read the contents of a sensitive file. This exposes all secrets to the agent's context and any associated logging systems. While necessary for the stated migration task, it constitutes a data exposure risk.
[COMMAND_EXECUTION] (MEDIUM): The skill generates commands like op item create that pass secret values as plaintext command-line arguments. This is a security anti-pattern as secrets can be leaked through shell history files (e.g., .bash_history), system monitoring tools, or process listings (ps).
[EXTERNAL_DOWNLOADS] (LOW): The skill installs the fnox tool via mise. While fnox is the central component of this skill, downloading and executing external binaries introduces a dependency on the integrity of the tool's distribution channel.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill processes external data from .env files.
Ingestion points: Reading .env file via cat command.
Boundary markers: Absent; the agent reads raw file content without delimiters.
Capability inventory: Subprocess execution (mise, op, git, rm).
Sanitization: Absent; the skill does not validate or sanitize the content of the .env file before processing.

env-to-fnox