skills/minhtri2710/skills/beo-author/Gen Agent Trust Hub

beo-author

Pass

Audited by Gen Agent Trust Hub on May 6, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [NO_CODE]: The skill consists entirely of Markdown files (SKILL.md, manual-pressure-scenarios.md, skill-writing-method.md) containing prose instructions and manual review scenarios. No executable scripts, binaries, or logic were found.
  • [SAFE]: The skill defines clear writable surface boundaries and includes hard stops to prevent modification of its own configuration or runtime artifacts without explicit user request. No malicious patterns such as credential harvesting or data exfiltration were detected.
  • [PROMPT_INJECTION]: The skill processes other skill definitions, representing an indirect prompt injection surface. However, it is an authoring tool and does not execute the content of the files it modifies.
  • Ingestion points: skills/beo/<skill>/SKILL.md and skills/beo/<skill>/references/*.md (as specified in SKILL.md Writable surfaces).
  • Boundary markers: Not present; the skill is designed for prose review of embedded content.
  • Capability inventory: Writable access to skill contracts and local writing guidance files as defined in SKILL.md.
  • Sanitization: Not present; the skill follows a manual review process for hardening wording.
Audit Metadata
Risk Level
SAFE
Analyzed
May 6, 2026, 04:07 AM
Security Audit — agent-trust-hub — beo-author