blame

Pass

Audited by Gen Agent Trust Hub on Jul 3, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill ingests untrusted data from git commit history via git log. It lacks explicit boundary markers or instructions to ignore embedded commands within the commit messages, creating a surface for indirect prompt injection.\n
  • Ingestion points: Git commit history and author names retrieved in Step 4.\n
  • Boundary markers: Absent; the instructions do not delimit the git log output from the agent's instructions.\n
  • Capability inventory: Execute (for git commands), Read, Grep, LS, AskUser.\n
  • Sanitization: Absent; the skill directly processes and narrates findings from the log.\n- [COMMAND_EXECUTION]: Uses the Execute tool to run standard git commands like git log and git config. These operations are consistent with the skill's purpose of analyzing repository history and are used within a controlled context.\n- [DATA_EXFILTRATION]: Provides a link for the user to manually share a report on X. This is a user-initiated action and does not involve automated or silent data exfiltration. References to author-controlled documentation at docs.factory.ai are included for user guidance.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 3, 2026, 10:07 AM
Security Audit — agent-trust-hub — blame