skills/eaedave/skills/business-readme/Gen Agent Trust Hub

business-readme

Pass

Audited by Gen Agent Trust Hub on Jul 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a bundled Node.js script scripts/validate-business-readme.cjs. This script validates the structure and marker tags within README.md and docs/LLM_CONTEXT.md to ensure documentation quality.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted project context (source code, existing documentation) to derive business rules.
  • Ingestion points: Project files including README.md, docs/, tests, and source code files (SKILL.md).
  • Boundary markers: Utilizes marker tags such as <!-- business-readme:business-rules:start --> to isolate and identify AI-managed content sections (SKILL.md).
  • Capability inventory: File system write access to specific documentation files and the ability to execute a local validation script (SKILL.md).
  • Sanitization: The instructions provide a strict output schema and human-readable focus, limiting the potential for malicious instructions in project data to influence the agent (SKILL.md).
  • [SAFE]: No network requests, external downloads, or access to sensitive system configuration files were detected. The skill operates exclusively on the local project directory provided in the context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 30, 2026, 01:31 PM
Security Audit — agent-trust-hub — business-readme