The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the lev task validate command to verify the structural integrity of generated task artifacts. This is a vendor-specific CLI tool used for internal validation within the development lifecycle.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from multiple sources to generate executable YAML files.
Ingestion points: Reads data from 'chat intent', repository 'evidence', and source design files (e.g., .lev/pm/designs/*.md).
Boundary markers: The skill uses XML-style delimiters like <transition-packet> and <final-proposal> to structure its output and separate instructions from data.
Capability inventory: The skill has the ability to write files to the .lev/pm/tasks/ directory and execute the lev CLI tool via subprocess.
Sanitization: There is no explicit sanitization or filtering of the external content before it is interpolated into the task DNA and execution plans.

propose