issue-summary
Warn
Audited by Gen Agent Trust Hub on Feb 15, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- PROMPT_INJECTION (MEDIUM): Vulnerable to Indirect Prompt Injection through external data.
- Ingestion points: The skill retrieves issue descriptions and comments via
gh issue viewin Phase 1. - Boundary markers: Absent. The skill lacks delimiters (e.g., XML tags or triple quotes) to separate the untrusted issue content from the agent's instructions.
- Capability inventory: The skill utilizes the
ghCLI for data retrieval. It does not appear to have write permissions or arbitrary code execution capabilities, which limits the impact of an injection to reasoning/summarization bias. - Sanitization: Absent. The external text is processed raw, allowing an attacker to place 'jailbreak' or 'ignore instructions' text within a GitHub comment to manipulate the summary output.
- COMMAND_EXECUTION (LOW): Uses the
gh(GitHub CLI) tool for its intended purpose. - The commands
gh issue vieware legitimate uses of the tool for the skill's described function.
Audit Metadata