international-expansion
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill features a vulnerability surface where user-supplied input is ingested and subsequently written to persistent files and briefing documents.
- Ingestion points: User input collected during the "Information gathering" phase (Step 1) in
SKILL.md(e.g., roles, country names, strategic context). - Boundary markers: None. The user input is directly interpolated into Markdown briefings and a YAML tracker without specific delimiters to distinguish between data and instructions.
- Capability inventory: The skill has permissions to read and write files within the
~/.claude/plugins/config/claude-for-legal-zh/directory. - Sanitization: There is no evidence of input validation or escaping for the strings being written to the file system or included in prompts.
- [SAFE]: File system activity is confined to a plugin-specific configuration directory (
~/.claude/plugins/config/claude-for-legal-zh/), which is standard behavior for maintaining project state in this environment. - [SAFE]: The skill does not attempt network connections, hardcode credentials, or execute arbitrary shell commands.
Audit Metadata