qa
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by processing untrusted user conversational input and local codebase files to generate GitHub issues.\n
- Ingestion points: User-provided bug descriptions and project files (e.g.,
UBIQUITOUS_LANGUAGE.md) are read into the agent's context inSKILL.md.\n - Boundary markers: There are no explicit delimiters or specific instructions for the agent to ignore potentially malicious instructions embedded within the user's reported bug or the codebase files.\n
- Capability inventory: The skill utilizes the GitHub CLI (
gh issue create) for external writes and invokes a subagent (subagent_type=Explore) for codebase analysis.\n - Sanitization: The agent is instructed to format the output using predefined templates, providing implicit structure, but there is no formal sanitization, validation, or escaping of the ingested external content.\n- [COMMAND_EXECUTION]: The skill executes external CLI commands to perform its primary function.\n
- Evidence: Uses
gh issue createto submit the gathered information to a remote repository. This is an intended capability but represents the tool's execution boundary.
Audit Metadata