The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes an external validation script located in a sibling skill directory.
Evidence: The instructions in SKILL.md call for running python3 <BOLD>à¸à¸à¸¥à¹à¸à¸´à¸£à¹ skill jt-next-step</BOLD>/gate.py reach-review --root ..
[COMMAND_EXECUTION]: The skill executes dynamic shell commands defined as placeholders within project data files.
Evidence: The skill processes and executes strings stored in variables like <TEST_CMD>, <SETUP_CMD>, <LINT_CMD>, <TYPECHECK_CMD>, and <BROWSER_TOOL> which are retrieved from the board.md or card files.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via the processing of untrusted content in task board files.
Ingestion points: card-loop/board.md and task card files located under card-loop/backlog/ in SKILL.md.
Boundary markers: No delimiters or specific safety warnings are used to prevent the agent from obeying instructions embedded within these data files.
Capability inventory: The skill can execute arbitrary shell commands, perform git operations (checkout, merge, push), and modify files within the repository (SKILL.md).
Sanitization: There is no evidence of validation or sanitization for the command strings (e.g., <TEST_CMD>) before they are passed to the shell for execution.

jt-do-work