om-auto-write-spec

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [SAFE]: No malicious patterns or security violations were detected. The skill is designed with a security-focused approach for handling external data and includes defensive mechanisms against privilege relaxation through local overrides.
  • [COMMAND_EXECUTION]: The skill performs shell-based operations for environment isolation using Git worktrees. These operations are restricted to the repository's temporary directories and follow standard practices for automated development workflows, including proper cleanup procedures.
  • [PROMPT_INJECTION]: The skill proactively addresses the risk of indirect prompt injection by defining a strict 'untrusted content boundary'. It instructs the agent to treat repository and tracker data (such as issues and PR bodies) exclusively as data, ignore embedded commands, and validate all externally-sourced values before shell or path interpolation. This includes explicit guards to prevent local configuration files from overriding safety or network rules.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 04:32 AM
Security Audit — agent-trust-hub — om-auto-write-spec