generating-sorbet-inline
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
bundle exec srb tc(the Sorbet type checker) to validate the generated signatures. It also includes a command to remove a temporary tracking file (rm .sorbet-inline-generation-todo.tmp). These are standard, low-risk operations within a development environment and are necessary for the skill's stated purpose. - [INDIRECT_PROMPT_INJECTION]: The skill represents an indirect prompt injection surface because it reads and processes user-provided Ruby source files, which could contain malicious instructions in comments or string literals.
- Ingestion points: Ruby source files are read during Step 1 (Analyze the Ruby Source) and Step 2 (Add Sorbet Signatures) as defined in
SKILL.md. - Boundary markers: The skill uses a checklist and a tracking file to structure its operations, but does not implement specific delimiters for untrusted code content.
- Capability inventory: The skill has file-write capabilities (updating Ruby files) and shell command execution (
rm,srb tc). - Sanitization: No specific sanitization or escaping of the Ruby code content is mentioned.
- Mitigation: The risk is significantly mitigated by a hard rule in the
SKILL.mdfrontmatter: "You MUST NOT run Ruby code of the project."
Audit Metadata