The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to run commands where the $ARGUMENTS variable is directly interpolated. This allows for command injection if a user provides a task ID containing shell control characters. Additionally, the same variable is used to construct file paths for reading and editing, which may be vulnerable to path traversal attacks (e.g., using ../ in the task ID).
[PROMPT_INJECTION]: The skill processes content from configuration and worklog files without validation, creating an indirect prompt injection surface.
Ingestion points: .taskmd.yaml and worklog markdown files (e.g., tasks/.worklogs/<ID>.md).
Boundary markers: None; there are no instructions to ignore or sanitize instructions found within these files.
Capability inventory: The skill can execute shell commands via Bash, and read or edit files using Read and Edit tools.
Sanitization: The skill does not validate the $ARGUMENTS variable or the content of the files it processes before using them in sensitive operations.

complete-task