turbo-transforms

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly exposes the agent to untrusted third-party content by providing a fetch_abi(url, format) SQL function (SKILL.md "Goldsky SQL Functions" — fetch_abi) that fetches ABI/IDL from arbitrary remote URLs and by documenting handler transforms that call external HTTP endpoints (references/typescript-transforms.md "Handler (External HTTP) Transforms"), both of which are read and used at runtime to drive decoding/enrichment logic.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill includes runtime fetches that can inject external data into pipeline logic—e.g., fetch_abi('https://example.com/erc20.json') is explicitly a runtime ABI fetch used by _gs_log_decode, and handler transforms call external endpoints like https://my-enrichment-api.example.com/process at runtime to enrich/transform records—both are runtime external dependencies that can directly control pipeline behavior.