The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill relies on bunx oh-my-ag to download and execute code directly from a remote package registry. This executes arbitrary logic from an unverified third-party source at runtime.
Evidence: Found in SKILL.md instructions: bunx oh-my-ag, bunx oh-my-ag doctor, bunx oh-my-ag dashboard.
[COMMAND_EXECUTION]: The skill performs multiple shell operations including spawning background processes and network bridges.
Evidence: bunx oh-my-ag dashboard & and bunx oh-my-ag bridge http://localhost:12341/sse in SKILL.md.
[EXTERNAL_DOWNLOADS]: The skill uses bunx to fetch external dependencies from the npm registry without prior verification or version pinning.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection and command injection because user-provided task descriptions are interpolated directly into shell commands used to spawn agents.
Ingestion points: Task descriptions passed to oh-my-ag agent:spawn (e.g., "Set up Unity3D build pipeline", "Implement auth API") in SKILL.md.
Boundary markers: None detected; instructions are passed as raw strings.
Capability inventory: The skill uses Bash, Write, and bunx which allows for arbitrary code execution if task strings contain shell metacharacters.
Sanitization: No evidence of input validation or escaping for the agent:spawn task arguments.

ohmg