siyu-restore

Pass

Audited by Gen Agent Trust Hub on Aug 10, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed to manage session state through local files and follows security best practices by explicitly prohibiting path traversal attacks.
  • [DATA_EXFILTRATION]: The skill restricts file operations to the ~/.siyu/clients/ directory, which is a vendor-specific path. No network exfiltration or unauthorized file access patterns were identified.
  • [INDIRECT_PROMPT_INJECTION]: The skill reads from local data archives, which constitutes a localized injection surface. 1. Ingestion points: Local markdown and frontmatter files in ~/.siyu/clients/ (SKILL.md). 2. Boundary markers: None implemented for file content interpolation. 3. Capability inventory: Summary generation and conditional triggering of other skills via the next_skill field (SKILL.md). 4. Sanitization: Path validation is present to ensure directory integrity, though specific content-level sanitization is not detailed.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 10, 2026, 02:03 AM
Security Audit — agent-trust-hub — siyu-restore