maintain-translated-skills
Fail
Audited by Gen Agent Trust Hub on Jul 21, 2026
Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands specified under the validationCommands field within a repository's policy.json file. This creates a direct path for arbitrary command injection if the agent processes a malicious or compromised repository.
- [REMOTE_CODE_EXECUTION]: The workflow involves running a Node.js validation script located at an external path. Executing code from paths external to the skill package increases the risk of running unauthorized or modified code.
- [PROMPT_INJECTION]: The skill processes external repository content and instructions from a policy file, creating a surface for indirect prompt injection that could influence the agent's file modification and command execution behavior.
Recommendations
- AI detected serious security threats
Audit Metadata