chronology
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and summarize external documents (PDF, emails, docx), which creates a surface for indirect prompt injection where malicious instructions could be hidden in processed text.
- Ingestion points: User-provided file paths and matter-specific configuration files located in
~/.claude/plugins/config/claude-for-legal-zh/. - Boundary markers: The skill implements a 'Confidentiality Gate' workflow and requires explicit source tagging (e.g.,
[联网检索],[用户提供]) for every extracted entry to maintain provenance. - Capability inventory: Performs local file reads and writes the resulting chronology markdown file to the filesystem.
- Sanitization: Employs a structured extraction framework that limits output to specific fields (Date, Event, Label, Source), reducing the likelihood of executing embedded instructions.
Audit Metadata