shopify-custom-data

SUSPICIOUS: the Shopify-focused guidance is broadly legitimate and install trust is low-risk, but the mandatory logging step routes the user's full prompt and correlation identifiers to Shopify on every use. That outbound telemetry is disproportionate to the stated documentation purpose, though same-org ownership, disclosure, and opt-out keep this below malicious.

No direct indicators of traditional malware behavior (no code execution primitives, filesystem manipulation, or suspicious process control) are present in this module. However, it clearly functions as a telemetry/usage reporter that transmits potentially sensitive user prompt content (truncated) and session identifiers over the network, and it may include a MINERVA_TOKEN as a Cookie header for staging requests. The telemetry destination can also be overridden via SHOPIFY_DEV_INSTRUMENTATION_URL, which increases the importance of strict allowlisting and secure configuration management. Overall: low likelihood of intentional malware, but moderate-to-high privacy and configuration-based security risk.

This module is primarily a telemetry/webhook sender that may include decoded (base64-derived) user prompt content in an outbound JSON payload and POST it to a configured remote endpoint. While the code avoids obvious injection issues (jq JSON escaping in the prompt path) and shows no overt malware/persistence/RCE patterns, it meaningfully increases privacy/exfiltration risk—especially because the network transmission is asynchronous and errors are suppressed, and because test mode can still leak the full request body (including user_prompt) to logs.