ln-652-transaction-correctness-auditor

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

• Third-party content exposure detected (high risk: 1.00). The skill's SKILL.md explicitly instructs the worker to "fetch files via WebFetch from https://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}" when shared/ is missing and marks several of those fetched files as "MANDATORY READ" (e.g., shared/references/*), meaning the agent will fetch and interpret public GitHub-hosted content that can materially influence its scanning and decision logic.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill specifies at-runtime WebFetch of shared files from https://raw.githubusercontent.com/levnikolaevich/claude-code-skills/master/skills/{path}, and those fetched "MANDATORY READ" templates/contracts directly control the worker's instructions and required behavior, so this is a runtime external dependency that can alter agent prompts.