de-aigc-skills

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided manuscripts, which introduces a surface for indirect prompt injection where instructions embedded in the input text could attempt to influence agent behavior.
  • Ingestion points: The skill receives external academic text in English and Chinese for auditing and rewriting, as documented in Step 0 of the SKILL.md file.
  • Boundary markers: The instructions mandate the creation of a structured 'AI-signature audit' report in table format in SKILL.md Step 1, providing a logical separation between the ingestion of text and the subsequent rewriting process.
  • Capability inventory: The skill functions as a set of instructions for the agent's internal reasoning and text generation; it does not utilize subprocesses, network requests, or file-writing tools.
  • Sanitization: No explicit input sanitization or validation logic is defined to filter potentially malicious instructions within the user-provided text.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 10:49 AM
Security Audit — agent-trust-hub — de-aigc-skills