The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a step that executes a bash command: gh issue create --title "$TITLE" $EXTRA_FLAGS. This command directly interpolates the $TITLE and $EXTRA_FLAGS variables, which are derived from user input. This creates a command injection vulnerability where a user could provide a title containing shell metacharacters (e.g., ;, |, or backticks) to execute unauthorized commands in the agent's environment.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection. Ingestion points: It reads the conversation context to generate a body for the GitHub issue if one is not provided. Boundary markers: There are no delimiters or instructions to ignore embedded commands within the ingested context. Capability inventory: The skill uses the gh command-line tool, providing write access to GitHub repositories. Sanitization: No sanitization or validation is applied to the conversation context before it is processed.

new-issue