Skills
skills/aaron-he-zhu/aaron-marketing-skills/dark-social-attributor/Gen Agent Trust Hub

dark-social-attributor

Pass

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from GA4/GSC exports and user surveys. Evidence: 1. Ingestion points: GA4/GSC exports, form inventory, survey answers in SKILL.md. 2. Boundary markers: Present (explicit instruction to treat as untrusted and ignore embedded commands). 3. Capability inventory: Execution of local python script and file writes to memory directories. 4. Sanitization: Relies on instruction-based filtering.
  • [COMMAND_EXECUTION]: Executes a local script scripts/connectors/pageviews.py to fetch Wikipedia pageview statistics. This operation is limited to a specific task using a local file and does not involve arbitrary user input.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 6, 2026, 03:38 PM
Security Audit — agent-trust-hub — dark-social-attributor