go-deploy
Fail
Audited by Gen Agent Trust Hub on Jun 25, 2026
Risk Level: HIGHCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: Phase 5 of the skill executes system shell commands including 'open', 'xdg-open', and 'cmd.exe /c start' using URL strings parsed from project files. These extracted URLs are interpolated directly into shell command templates.
- [PROMPT_INJECTION]: The skill processes untrusted data from various project files which can influence subsequent shell command execution.
- Ingestion points: Files scanned include 'docs/deploy.md', 'AGENTS.md', 'README.md', the 'scripts/' directory, and CI/CD configuration files like '.github/workflows/*.yml', '.gitlab-ci.yml', and 'Jenkinsfile'.
- Boundary markers: No delimiters or instructions are used to separate untrusted file content from the agent's internal logic or command templates.
- Capability inventory: The skill is capable of reading project files, writing to the 'docs/' directory, and executing OS commands.
- Sanitization: The skill verifies that URLs start with 'http://' or 'https://' but does not perform escaping of shell metacharacters (e.g., semicolons, backticks, or subshell syntax) or validation against command injection payloads.
Recommendations
- AI detected serious security threats
Audit Metadata