skill-optimizer

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt instructs the agent to scan and quote raw user messages from local session transcripts as "evidence," which forces inclusion of verbatim user content (potentially API keys, tokens, passwords, or other secrets) in the generated report.

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

• Third-party content exposure detected (medium risk: 0.65). Outsider free text is ingested via Step 2: Collect Session Data, where the runtime scans and reads session JSONL transcripts from ~/.claude/projects/**/*.jsonl / ~/.codex/sessions/**/*.jsonl and extracts full user messages (which may include third-party authored content) into the analysis pipeline/LLM context.