The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied tasks and reads project files (such as code, data, and logs) to formulate its "ambition" plan, creating a surface for potential injection.
Ingestion points: The user-provided <task> or <prompt> arguments, and any project files read during the baseline measurement phase (SKILL.md).
Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded directives in the data it processes.
Capability inventory: The agent reads project files, executes a local helper script (deal.py), and writes a planning brief (BRIEF.md) to the filesystem.
Sanitization: No explicit sanitization, escaping, or filtering of external content is specified before interpolation into the agent's context.
[COMMAND_EXECUTION]: The skill requires the agent to execute a local Python script (scripts/deal.py) to select appropriate mental models and Range fields for a given task.
The execution uses a relative path to a script distributed with the skill and relies only on the Python standard library.
A subprocess.run call is used within the test suite (tests/test_deal.py) to validate CLI behavior with static, controlled inputs.

ambition