The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script located at ${CLAUDE_SKILL_DIR}/../../scripts/new-feature.sh using a user-provided string (<description>) as a direct argument. This creates a potential command injection vector if the platform or the shell call does not properly escape characters in the description field.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted feature requests to generate technical specifications (spec.md).
Ingestion points: User-provided feature requests processed in SKILL.md steps.
Boundary markers: The skill uses a specific markdown template and mandatory [NEEDS CLARIFICATION] markers to isolate unknowns, which provides some structural defense.
Capability inventory: The skill has Write, Edit, and Bash (mkdir/ls) capabilities.
Sanitization: There is no explicit instruction to sanitize or escape the user-provided content before writing it to files or passing it to the new-feature.sh script.

specify