health
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes localized Bash and PowerShell scripts (e.g.,
collect-data.sh,run-health.ps1) to perform system health checks, context audits, and maintainability scans. These scripts are invoked with path sanitization to prevent the execution of untrusted binaries from the project root. - [INDIRECT_PROMPT_INJECTION]: As an auditing tool, the skill is designed to ingest untrusted project documentation (AGENTS.md, CLAUDE.md) and session history. This creates a surface for indirect prompt injection, which the skill mitigates by providing the agent with explicit instructions to treat all audited content as untrusted data rather than authoritative instructions.
- [PRIVILEGE_ESCALATION]: The skill utilizes the PowerShell
ExecutionPolicy Bypassflag to allow the execution of its local launcher scripts on Windows environments. This is scoped strictly to the audit process and does not modify the persistent system-wide execution policy. - [DATA_EXFILTRATION]: The skill accesses project configuration files and session logs to perform audits. This behavior is transparent and includes robust redaction logic in the included Python scripts to prevent the exposure of secrets, API tokens, or private keys during the reporting phase.
Audit Metadata