wenqu-review

Pass

Audited by Gen Agent Trust Hub on Aug 5, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs automated analysis of the local filesystem using standard utilities. It uses Bash grep to extract and audit English terminology (documented in references/r1-english.md) and utilizes git diff to identify content changes and ensure consistency across the document structure (documented in references/r7-change-closure.md). Additionally, the fact-checking stage (references/r0-factcheck.md) instructs the agent to verify technical claims against source code, configuration files, and execution records.
  • [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it is designed to ingest and act upon data from untrusted external articles and reports.
  • Ingestion points: External content is read from project files specified by the user or identified during the review process (e.g., in SKILL.md and references/r5-coherence.md).
  • Boundary markers: The instructions do not implement protective delimiters or specific 'ignore' directives to encapsulate the ingested text.
  • Capability inventory: The agent possesses Read, Write, and Edit permissions for project files and can execute shell-based search and version control commands.
  • Sanitization: There is no evidence of automated sanitization or filtering applied to the content of the articles before they are processed by the agent's logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 5, 2026, 08:31 AM
Security Audit — agent-trust-hub — wenqu-review