scienceclaw-summarization

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill consists exclusively of markdown-formatted instructions and templates. No executable scripts, binaries, or shell commands are present.
  • [SAFE]: Includes a 'Zero-Hallucination Rule' that forces the agent to rely only on tool results and prevents the fabrication of scientific data or citations, which is a defensive measure for information integrity.
  • [SAFE]: The skill provides structured guidance for discipline-specific terminology and citation handling, following standard academic practices.
  • [SAFE]: Analysis of Category 8 (Indirect Prompt Injection) surface: (1) Ingestion points: Processes scientific papers and literature (SKILL.md); (2) Boundary markers: Absent, the skill uses structured templates for output organization; (3) Capability inventory: References the use of tools (API, code, search) for data retrieval; (4) Sanitization: Not specified. The risk is minimized by the skill's focus on neutral, faithful summarization and traceability.
  • [SAFE]: No evidence of prompt injection, data exfiltration, or obfuscation (Base64, hex, or hidden characters) was detected in the content or metadata.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:31 PM
Security Audit — agent-trust-hub — scienceclaw-summarization