frontend-craft

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied architecture descriptions to facilitate routing.
  • Ingestion points: User technical queries describing frontend scenarios (as evidenced in evals/evals.json) enter the agent context.
  • Boundary markers: The instructions do not define explicit delimiters to isolate user-provided content from internal skill logic.
  • Capability inventory: Authorized tools in SKILL.md include file system operations (Read, Grep, Glob), a task execution tool (Task), and specialized documentation lookup tools (mcp__context7__resolve-library-id, mcp__context7__query-docs).
  • Sanitization: There is no evidence of input validation or sanitization for the architecture descriptions provided by users.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 02:32 PM
Security Audit — agent-trust-hub — frontend-craft