server-side-conversion-tracking

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches a docker-compose.yml configuration file from the author's repository on GitHub (github.com/autonnel/autonnel) to initialize the tracking infrastructure.
  • [COMMAND_EXECUTION]: The instructions include running docker compose up to start the downloaded service containers.
  • [DATA_EXFILTRATION]: As part of its intended purpose, the skill transmits conversion events and customer identifiers to external ad platforms (Facebook, TikTok, Google, and Bing) using their respective APIs.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from URL parameters (such as fbclid or gclid) and visitor metadata.
  • Ingestion points: URL query parameters and forwarded HTTP headers.
  • Boundary markers: None explicitly defined in the instructions.
  • Capability inventory: Network operations to ad platform APIs.
  • Sanitization: Employs SHA-256 hashing for PII (email, phone) as required by the ad platforms.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 01:56 AM
Security Audit — agent-trust-hub — server-side-conversion-tracking