Intent-Code Divergence
Medium
- Confidence
- 97% confidence
- Finding
- The skill claims robots.txt adherence, but on fetch errors or parse failures it defaults to allowing scraping. That fail-open behavior can cause the agent to access paths that a site intended to restrict, especially during transient network issues or malformed robots files. In a scraping skill, this makes the mismatch between policy claims and implementation materially relevant.
