solidity-auditor

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 1.00). The skill explicitly instructs bundling the entire codebase and delivering it to spawned agent "models" (sonnet/opus) and to fetch external URLs, creating an intentional data-exfiltration vector to external services / models.

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

• Third-party content exposure detected (low risk: 0.10). Outsider-authored free text is ingested indirectly only through user-controlled/local project files: the workflow uses Bash find to discover in-scope .sol files and concatenates their contents into /tmp/audit-agent-{1..4}-bundle.md for LLM reading; it also reads a local references/.../attack-vectors-*.md and possibly assets/findings/*.md (previous reports) only from the project’s own filesystem context, not from an external feed/queue.