skills/afelipeg/anthropic-skills-for-enterprise-marketing-os/weekly-control-tower/Gen Agent Trust Hub
weekly-control-tower
Pass
Audited by Gen Agent Trust Hub on Jun 14, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill adheres to its stated functional purpose of providing an operational intelligence layer for client health reviews.
- [DATA_EXFILTRATION]: The skill ingests data from MCP-connected platforms (e.g., Supermetrics, GA4) and user-uploaded files (CSV, Excel, PDF) as described in SKILL.md. No unauthorized network exfiltration, hardcoded credentials, or sensitive file path access was found in scripts/control_tower_engine.py or the instructions.
- [REMOTE_CODE_EXECUTION]: No remote code execution patterns, piped shell commands, or unverifiable dependencies were detected. The Python script scripts/control_tower_engine.py uses only standard library modules (json, math, dataclasses) to perform local calculations.
- [PROMPT_INJECTION]: The skill was evaluated for indirect prompt injection risk (Category 8) due to its ingestion of untrusted external data. Ingestion points: MCP platform data and uploaded reports (CSV, Excel, PDF). Boundary markers: Absent. Capability inventory: Data summarization and visual rendering; no dangerous exec/eval functions in the processing script. Sanitization: Not explicitly present. The risk level is low and consistent with the intended primary purpose of the skill.
Audit Metadata