mermaid
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill mentions installation commands (
npx skills add) for related diagramming tools in theSKILL.mdrouting section. These tools are authored by the same vendor (full-stack-skills) and the instructions are intended for legitimate agent setup.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided source files to generate diagrams, creating an inherent attack surface. However, the functionality is restricted to diagram code generation.\n - Ingestion points: Workflow step 1 in
SKILL.mdinvolves extracting entities and states from user requests and available source code.\n - Boundary markers: The instructions do not define specific delimiters for separating untrusted source data from internal logic.\n
- Capability inventory: The skill can generate text-based diagram source and save Markdown files to the local repository (e.g.,
docs/diagrams/).\n - Sanitization: Input data is interpolated into diagram labels and structures without explicit sanitization steps mentioned in the markdown.
Audit Metadata