nda-review-jamie-tso
Pass
Audited by Gen Agent Trust Hub on Oct 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted legal documents, which establishes an inherent surface for indirect prompt injection attacks.
- Ingestion points: The skill processes user-supplied NDA text during the triage and clause-by-clause review phases, as outlined in
SKILL.mdand supported by the reference modules in thereferences/directory. - Boundary markers: The playbook instructions do not include requirements for the agent to use explicit delimiters or boundary markers (such as XML tags or unique string wrappers) to isolate the untrusted document content from the agent's logic.
- Capability inventory: The skill's capabilities are restricted to textual analysis, reasoning, and generating formatted Markdown reports. It lacks capabilities for network access, file system modifications, or subprocess execution.
- Sanitization: The workflow does not provide instructions for the agent to sanitize, filter, or escape potential instructions hidden within the source NDA text that might attempt to influence the agent's behavior.
Audit Metadata