learn-rat-stack
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The instructions include a final verification step using
pnpm turbo run check test build. This is a standard command pattern for linting, testing, and building modern TypeScript projects using the Turbo build system. - [DYNAMIC_EXECUTION]: The skill documents a 'code mode' feature that enables the execution of JavaScript logic via a hosted API endpoint (
https://ratstack.sh/api/execute). This is explicitly presented as a feature for agents to compose multiple tool calls into a single program. - [INDIRECT_PROMPT_INJECTION]: The skill describes tools for searching and reading hosted content, which defines an ingestion surface for untrusted external data. While the skill doesn't implement the sanitization itself, it provides a structured 'contract' system to define input and output schemas for these capabilities.
Audit Metadata