paper-polish

Pass

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or security vulnerabilities were identified within the skill files or instructions.
  • [NO_CODE]: The skill is entirely comprised of Markdown files. It does not include any Python scripts, Node.js modules, binaries, or shell commands.
  • [PROMPT_INJECTION]: While the skill processes untrusted data (user-supplied manuscripts), it does so within a strictly defined instructional framework.
  • Ingestion points: Academic drafts provided by the user via paste or workspace files (SKILL.md).
  • Boundary markers: None identified.
  • Capability inventory: The agent is instructed to read and edit Markdown or LaTeX files in the workspace (SKILL.md).
  • Sanitization: No specific sanitization is mentioned; however, the instructions include extensive guardrails requiring the agent to remain faithful to the original meaning and explicitly flag any potential changes to scientific intent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 10, 2026, 01:23 PM
Security Audit — agent-trust-hub — paper-polish