daily-review

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Prompt Injection (HIGH): The skill is susceptible to Indirect Prompt Injection (Category 8) due to the processing of external data. \n
  • Ingestion points: GitHub commit messages are retrieved via the gh search commits command in Step 3. \n
  • Boundary markers: Absent. The instructions do not specify any delimiters or safety warnings to distinguish between user-provided data and system instructions when processing the commit history. \n
  • Capability inventory: The skill is granted Write, Edit, and Bash(gh:*) permissions. This combination allows an attacker who successfully injects instructions into a commit message to potentially modify local files, overwrite the journal, or perform unauthorized GitHub operations. \n
  • Sanitization: Absent. There is no logic provided to sanitize or filter the content of commit messages before the agent summarizes them. \n
  • Command Execution (MEDIUM): The skill requests the Bash(gh:*) tool. While intended for commit searches, the broad glob pattern gh:* could allow the agent to execute any GitHub CLI command (e.g., deleting repositories or managing secrets) if tricked by an injection.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 05:03 AM