The Agent Skills Directory

[PROMPT_INJECTION]: The skill relies on a variable placeholder {feature-name} which is interpolated into bash scripts. If the provided feature name contains shell metacharacters (e.g., ;, &&, or ||), it could allow for arbitrary command execution beyond the intended grep and find operations.
[COMMAND_EXECUTION]: The skill utilizes local system utilities including grep, find, ls, and wc to perform file system analysis. While these are necessary for the skill's purpose, their use with un-sanitized input strings creates a security risk surface.
[PROMPT_INJECTION]: Evidence Chain for Indirect Prompt Injection:
Ingestion points: The {feature-name} parameter in the 'Quick Evaluation Script' section of SKILL.md serves as the entry point for untrusted data.
Boundary markers: Absent. There are no delimiters or instructions to treat the input as a literal string or to sanitize shell-specific characters.
Capability inventory: The skill has the capability to execute shell commands via the agent's environment, specifically targeting file reads and directory listing across the project source tree.
Sanitization: Absent. The variable is used via direct string interpolation in the shell script block.

gs-evaluate-domain-module