Credential Exfiltration from AI Coding Assistants via Indirect Prompt Injection in Fetched Web Content | AI Breaking Wire