legal-risk-assessment-anthropic
Warn
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill's YAML frontmatter identifies the author as 'Anthropic', which contradicts the provided author context of 'lawve-ai'. This use of a high-profile organization's name is a deceptive practice intended to gain unearned trust.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data concerning legal matters and contracts, which presents a surface for potential indirect prompt injection.
- Ingestion points: Sections for 'Matter' and 'Background and Context' in the Risk Assessment Memo template defined in SKILL.md.
- Boundary markers: No explicit delimiters or instructions to ignore instructions found within user-provided text are present in the skill instructions.
- Capability inventory: This skill does not use any executable tools, shell commands, file system access, or network operations.
- Sanitization: No sanitization or content validation logic is specified in the instruction files.
Audit Metadata