agent-friendly

Pass

Audited by Gen Agent Trust Hub on Nov 21, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to autonomously execute scripts defined within the repository being audited, specifically check, test, doctor, and setup.sh. In 'Build' and 'Apply' modes, these are run without user confirmation to achieve a 'green' state. In 'Audit' mode, they are run 'freely'. This poses a significant risk if the agent is directed to audit a malicious or untrusted repository, as these scripts could execute arbitrary commands on the host system.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process instructions and source code from external repositories to guide its auditing and improvement tasks. This creates a surface for indirect prompt injection where malicious instructions embedded in a repo's AGENTS.md, LAWS.md, or code comments could override the agent's behavior or cause it to perform unintended actions.\n
  • Ingestion points: Repository source files, instruction files (AGENTS.md, LAWS.md, REVIEW.md), and various configuration registries found in the codebase being audited.\n
  • Boundary markers: The skill does not define explicit delimiters or 'ignore' instructions for the content it reads from the audited repositories.\n
  • Capability inventory: The skill has broad capabilities including file system writes (Build and Apply modes), script execution (check, test, setup.sh), and potentially network access depending on the repository's own tools.\n
  • Sanitization: There are no instructions for sanitizing, escaping, or validating the content of the files read from the repository before the agent incorporates them into its reasoning or uses them to determine actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 21, 2026, 03:50 AM