cs-health-scorecard
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes a local Python script,
scripts/health_score.py, to perform weighted scoring calculations. This script is part of the skill package and relies exclusively on the Python standard library. \n- [DATA_EXFILTRATION]: Accesses sensitive account information (stakeholders, contract values, usage data) stored in a local 'brain' repository viabrain_query.pyand updates it viabrain_write.py. This data handling is appropriate for the skill's intended purpose in account management. \n- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection. \n - Ingestion points: Data enters the context via user input prompts and the output of
brain_query.pyinSKILL.md. \n - Boundary markers: None identified; untrusted data is directly interpolated into markdown templates. \n
- Capability inventory: Subprocess execution of local scripts (
health_score.py) and filesystem updates via thebrain_write.pyhelper. \n - Sanitization: None identified; no escaping or filtering is applied to the ingested account data before processing.
Audit Metadata