The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates an interview process where user-provided requirements are written directly into a file (card-loop/brief.md). This establishes an indirect prompt injection surface where malicious instructions could be embedded in the brief and later parsed or executed by other agents or tools.
Ingestion points: User input provided during the requirement-gathering interview (SKILL.md).
Boundary markers: The skill does not define or use boundary markers or delimiters to isolate user-provided content within the generated brief.md file.
Capability inventory: The skill utilizes file writing capabilities to create card-loop/brief.md and instructs the agent to perform Git operations (checkout, commit).
Sanitization: There is no evidence of input validation, sanitization, or escaping of the user's responses before they are written to the filesystem and committed to the repository.
[COMMAND_EXECUTION]: The skill directs the agent to perform repository management tasks that depend on data sourced from the environment.
Evidence: The "สร้าง" section directs the agent to checkout a branch based on a specific line ("สาขาคิว:") found in the local context (likely card-loop/board.md). If the source file containing this branch name is modified by an attacker, it could lead to the agent executing unintended Git commands or switching to unauthorized branches.

jt-ask-brief