hyper-report
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill defines a rendering pipeline in
references/render-pipeline.mdthat executes shell commands. It uses standard utilities such asawk,sed,grep,jq, andmkdirto manage document metadata and assets. It also invokes a Chromium-family browser for headless rendering tasks. - [DYNAMIC_EXECUTION]: Document layout and pagination are managed dynamically at runtime. The skill assembles HTML and executes a JavaScript runtime (
assets/runtime/report.js) within a headless browser to calculate page geometry and resolve cross-references. - [INDIRECT_PROMPT_INJECTION]: As a report generation tool, the skill is designed to ingest untrusted data from external sources and user-provided Markdown. It contains explicit safety policies in its
instruction_contractandforbiddensections, stating that retrieved content must be treated as evidence rather than instructions, which mitigates the risk of the agent obeying malicious commands embedded in source documents.
Audit Metadata