library-cleanup

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The skill requires reading file contents with cat_file and then embedding content-derived summaries and proposed filenames into reports/checklists/dashboards, which can force the LLM to output sensitive strings (API keys, passwords, tokens) verbatim if those appear in files.

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

• Third-party content exposure detected (high risk: 0.70). This skill’s workflow explicitly reads file content from the target “document library” (e.g., via cat_file in Phase 2 for badly-named files), so if those documents were authored by others, their free-form document text becomes LLM-ingested context for rename/categorization.