soia-dev-github-ops

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted content from GitHub Pull Request diffs and repository-specific rule files (e.g., CLAUDE.md, CONTRIBUTING.md). An attacker could embed instructions in these files to manipulate the agent's review verdict or trick it into performing unauthorized actions.
  • Ingestion points: The skill pulls data via 'gh pr diff' and 'gh api' for file contents from target repositories.
  • Boundary markers: While the skill instructs the agent to separate facts from inferences, it lacks explicit delimiters or instructions to ignore embedded commands within the ingested content.
  • Capability inventory: The skill possesses high-impact capabilities including 'gh pr merge', 'gh release create', and the ability to modify repository permissions via 'gh api -X PUT'.
  • Sanitization: There are no instructions for sanitizing, escaping, or filtering the external content before it is processed by the agent.
  • [COMMAND_EXECUTION]: The skill relies on executing system commands via the gh CLI to perform its functions.
  • High-impact operations include merging pull requests and modifying repository collaborator permissions.
  • The skill incorporates a 'Safety Model' and 'Safety Gate' section requiring explicit user confirmation for these sensitive operations to mitigate the risk of accidental or unauthorized execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 10:18 AM
Security Audit — agent-trust-hub — soia-dev-github-ops