claude-code-auto-advisor

Pass

Audited by Gen Agent Trust Hub on Jul 15, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill follows best practices for tool configuration and documentation. No malicious patterns, obfuscation, or persistence mechanisms were detected.
  • [COMMAND_EXECUTION]: Diagnostic Python scripts are included to check local Claude Code settings files for advisor model configuration. These scripts are restricted to read-only operations on application-specific configuration paths.
  • [EXTERNAL_DOWNLOADS]: Installation instructions use the standard npx command for the skill's ecosystem. The skill also references official documentation from trusted Anthropic domains.
  • [PROMPT_INJECTION]: The skill manages a workflow that ingests local data (files, logs, diffs) to provide context to an advisor model, creating a potential surface for indirect prompt injection. Ingestion points: Files, diffs, logs, and documentation defined in SKILL.md. Boundary markers: None specified. Capability inventory: Use of the advisor tool and execution of local configuration scripts. Sanitization: No explicit content filtering or sanitization is defined for the gathered context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 15, 2026, 10:32 AM
Security Audit — agent-trust-hub — claude-code-auto-advisor