Single tool 'phiscrub_scan' has a reasonable description (127 chars, within 10-1024 range) but critical gaps in schema completeness and parameter documentation. Input schema exists but lacks type validation details. No output schema documented. Tool name is verb-noun compliant ('scan' action). No error handling guidance, no parameter constraints, no security considerations for file access. The description mentions multiple PHI types (names, MRNs, SSNs, dates, addresses) but does not explain what the tool returns or when to use it vs alternatives.

• Document the complete output schema: e.g. {findings: [{id, severity, type, location, value, redacted_value}], scan_summary: {total_findings, files_scanned, errors}}. Include field types and descriptions.
• Add parameter constraints to 'target': specify accepted formats (absolute/relative path), file type filters (.log, .csv, .txt), and max directory depth. Example: 'target (string, required): File or directory path (max 1000 chars, no symlinks). Scans .log, .csv, .txt files recursively up to 10 levels.'
• Add error handling guidance: 'Returns error if path does not exist, is not readable, or exceeds size limits (>1GB). If permission denied, check file ownership and agent credentials.'
• Clarify redaction behavior: 'If redact=true, modifies files in place (irreversible). If redact=false, returns findings without modifying originals. Recommend dry-run first.' Add a 'dry_run' boolean parameter (default true) to prevent accidental data loss.
• Add security note: 'Requires read access to target path. Audit logs all scans with timestamp, user, path, and findings count. Does not transmit PHI outside the local system.'