tool-evaluator
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external tool data, vendor claims, and user scenarios, creating an attack surface for instructions embedded within the processed content.\n
- Ingestion points: Ingests tool specifications via
tool_config,required_features, andoptional_featureswithin theSKILL.mdcode and workflow.\n - Boundary markers: The instructions lack explicit delimitation or warnings to ignore malicious instructions embedded in external data.\n
- Capability inventory: The skill possesses network access capabilities through the Python
requestslibrary.\n - Sanitization: There is no evidence of validation or sanitization of input data before it is processed or used in reporting.\n- [DATA_EXFILTRATION]: The skill includes functionality for network communication that could be targeted at arbitrary endpoints.\n
- Evidence: The
_test_performancemethod inSKILL.mdperformsrequests.get(api_endpoint)where the target is defined by the user in the tool configuration. This allows for potential network probing or interaction with internal services (SSRF).
Audit Metadata