multi-agent-testing
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill includes local shell and Python scripts (
evals/fixtures/check-domain-decisions.sh,evals/fixtures/check_domain_decisions.py) used for validating test outputs. These scripts are confined to the local environment and use standard libraries for regex-based table parsing and data comparison. They do not perform network operations or access sensitive system files. - [INDIRECT_PROMPT_INJECTION]: The skill incorporates robust defensive instructions to mitigate risks from untrusted input. The
prompts/multi-agent-testing.mdrequires an 'Input Audit' to explicitly identify missing, conflicting, or out-of-scope information. It also mandates the use of least-privilege, masked data, and dry runs when handling user or production data. - [DATA_EXFILTRATION]: No network operations or hardcoded credentials were found. The skill operates on local references and does not attempt to send data to external endpoints.
- [DYNAMIC_EXECUTION]: While the skill package contains executable scripts for evaluation purposes, they are static and do not involve runtime code generation from untrusted sources. The Python script performs deterministic checks against a hardcoded oracle.
Audit Metadata