tune

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.95). The skill requires quoting transcript lines and writing exact diffs/edits after reading local transcripts and settings files (including settings.local.json), so any secrets present there would be copied verbatim into the skill's outputs/files, creating an exfiltration risk.

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

• Third-party content exposure detected (medium risk: 0.30). This skill reads locally stored transcript JSONL evidence from ~/.claude/projects/<project-dir>/*.jsonl, plus local first-party project files, and mines user-authored correction/prose at runtime rather than requiring an explicit targeted fetch.