astrofy-mdx
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion and processing of MDX files, which are capable of executing JSX code. This presents an inherent attack surface for indirect prompt injection if the agent processes untrusted content.\n
- Ingestion points: The skill instructions direct the agent to read and compile MDX and Markdown files from
src/content/blogor other project-specific paths defined in.astrofy/config/paths.json.\n - Boundary markers: The skill specifically instructs the user to limit MDX imports to components approved by the project and to use Zod schemas to validate frontmatter, which acts as a boundary against malformed or malicious metadata.\n
- Capability inventory: The process involves MDX compilation (JSX execution) and the use of the
astrofyCLI tool for verification.\n - Sanitization: The documentation explicitly warns about the risks of MDX and advises treating third-party content as plain Markdown without execution to prevent arbitrary code imports.\n- [COMMAND_EXECUTION]: The skill utilizes a vendor-specific CLI tool,
astrofy, for project verification and diagnostics tasks (e.g.,astrofy check --category mdx).
Audit Metadata