contract-reviewer
Pass
Audited by Gen Agent Trust Hub on Apr 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script,
contract_reviewer.py, to manage document processing and API communication. - [DATA_EXFILTRATION]: Documents are sent to the vendor's domain
https://somark.techfor parsing. This is the primary function of the skill and uses the author's infrastructure. - [PROMPT_INJECTION]: The skill processes untrusted external legal documents, which serves as an attack surface for indirect prompt injection. 1. Ingestion points: Files are ingested via the
-fflag in thecontract_reviewer.pyscript. 2. Boundary markers: TheSKILL.mdfile explicitly instructs the agent to treat all parsed contract content as data and avoid executing any instructions found within it. 3. Capability inventory: The skill has the capability to read local files, write parsed outputs to disk, and communicate with the vendor's API. 4. Sanitization: The skill relies on natural language boundary instructions for the agent rather than programmatic content sanitization.
Audit Metadata