permix-getting-started
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the user or agent to install the
permixlibrary using standard package managers (pnpm,npm, oryarn). This is a legitimate operation for the skill's purpose of setting up the Permix library. - [INDIRECT_PROMPT_INJECTION]: The skill provides templates and configuration methods that ingest external data to determine permissions (e.g., resource data or user objects). This creates a surface where the logic could be influenced by untrusted data processed at runtime.
- Ingestion points: Data ingestion occurs in the
permix.setupandpermix.templatecallback functions defined inSKILL.md(e.g.,(post) => post.authorId === user.id). - Boundary markers: No explicit boundary markers or delimiters are suggested in the code examples to separate data from control logic.
- Capability inventory: The skill uses file modification and package installation capabilities to set up the library.
- Sanitization: The provided examples do not include explicit sanitization or validation of the input data objects before they are used in permission checks.
- Ingestion points: Data ingestion occurs in the
Audit Metadata