The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is explicitly designed to generate and execute shell commands using the Bash tool. Its core 'Iron Law' mandates that every planning task must include an executable command, creating a direct path from plan generation to command execution.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by transforming arbitrary user requirements into executable bash commands. A malicious user could provide requirements that trick the agent into generating harmful commands (e.g., data deletion or unauthorized access) as part of a legitimate-looking plan.
Ingestion points: User requirements parsed in the 'Analyze Requirements' step (Step 1).
Boundary markers: None present; the skill lacks delimiters or instructions to ignore embedded commands within user requirements.
Capability inventory: Access to Bash, Write, Edit, Glob, Grep, and Skill tools enables a wide range of file and system operations.
Sanitization: No validation or sanitization logic is implemented to filter potentially malicious requirements before they are converted into executable tasks.

plan-generator