The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes an unvalidated <statement to analyze> from the user. An attacker could craft a statement designed to influence the parallel agents' behavior or extract system information.
Ingestion points: The argument-hint field in SKILL.md defines the entry point for the user-provided statement.
Boundary markers: Absent. No delimiters or instructions to ignore embedded commands are used when the statement is passed to the agents.
Capability inventory: The skill utilizes Bash, Read, Grep, Glob, and Task tools, which provide extensive file system and execution permissions.
Sanitization: Absent. No validation or escaping is performed on the input statement before it is interpolated into the agent tasks.
[COMMAND_EXECUTION]: The skill explicitly authorizes the use of the Bash tool. In conjunction with the indirect prompt injection surface, this creates a risk where an agent might be manipulated into executing malicious commands found within the statement being analyzed.

dialectic