llms-txt-checker

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

• Third-party content exposure detected (high risk: 0.95). The skill uses bash_tool + curl to fetch outsider-authored free text from the user-supplied domain’s well-known paths (/robots.txt, /llms.txt, /llms-full.txt), then cat/head/wc those files into the agent’s runtime context for auditing and reporting.

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

• Potentially malicious external URL detected (high risk: 0.90). The skill uses bash_tool to fetch and inject remote content at runtime (curl "https://$DOMAIN/robots.txt", "https://$DOMAIN/llms.txt", and "https://$DOMAIN/llms-full.txt" and any custom paths discovered) into the model's context to drive its audit, so those external files can directly influence the agent's prompts/output.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)