diagnostic-issue-resolver

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

• Third-party content exposure detected (high risk: 0.80). The skill's required automated diagnostics explicitly tail/grep the Telegram audit NDJSON logs (~/.claude/automation/claude-telegram-sync/logs/audit/*.ndjson) and also probes public endpoints (e.g., curl https://api.telegram.org/ and HuggingFace model cache), meaning it ingests untrusted, user-generated Telegram/log content as part of its decision workflow and that content can materially influence diagnostics and fixes.