The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gh CLI to create issues, incorporating a user-derived title directly into the command line (gh issue create --title "<title>"). A malicious user could attempt to include shell metacharacters in the title to execute unauthorized commands if the executing agent fails to sanitize the input.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user descriptions to draft GitHub issues. Ingestion points: User-provided bug description. Boundary markers: The skill instructs the agent to use Markdown fencing for stack traces but does not specify delimiters for the overall input description. Capability inventory: Bash(gh *), Bash(git *) (file creation, issue creation). Sanitization: No specific sanitization or escaping of the user description is mentioned before it is structured into the Markdown report.

file-bug