quick-fix-suggestion

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill reads repository files and inserts "existing_code" verbatim into suggested code snippets (and uses file/content search), so any secrets present in those files could be output unchanged—creating an exfiltration risk.

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and analyzes Gitee issue details and comments using get_repo_issue_detail and list_issue_comments (SKILL.md Step 1 and the Tool Reference), ingesting untrusted user-generated issue/comment content that the agent interprets and uses to decide code changes and next actions.