start-execution-model
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides educational content on environment boundaries and secret management in TanStack Start, which is a security-enhancing practice to prevent data leakage.
- [COMMAND_EXECUTION]: The skill includes a mandatory workflow in
SKILL.mdrequiring the agent to execute thefuse-ai-pilot:snipertool, which is a platform-specific vendor utility. - [DYNAMIC_EXECUTION]: Code templates in
references/templates/boundaries.mddemonstrate standard environment-specific dependency loading usingrequire('node:fs')inside server blocks. - [INDIRECT_PROMPT_INJECTION]: The skill documents patterns for data flow through application boundaries.
- Ingestion points: Examples in
references/templates/boundaries.mduse loaders and functions to process external data. - Boundary markers: The skill instructs on isolating logic using
createServerFn,createClientOnlyFn, and file-level markers. - Capability inventory: Templates demonstrate capabilities including network access (
fetch) and file system access (node:fs). - Sanitization: Instructions mandate that secrets and sensitive operations be restricted to server-only functions to prevent client-side exposure.
Audit Metadata