VaultArchive has clear, action-verb naming (archive_zip, archive_list, archive_extract) and reasonable descriptions (avg ~80 chars). All 5 tools have input schemas with Zod validation and type definitions. However, parameter descriptions are minimal (often single phrases like 'Absolute paths of files to add to the zip'), lacking guidance on constraints, formats, or error recovery. Output schemas are not documented, callers must infer return structure from implementation. Error handling returns plain text messages without recovery guidance. No tool annotations (readOnlyHint/destructiveHint) despite clear risk profiles (WRITE vs READ_ONLY). Missing output schema documentation is a critical gap for LLM planning.

• Document output schemas for all tools. E.g., archive_list should return {entries: [{name: string, size_bytes: number}], total_count: number} instead of a formatted string. Structured output enables LLM planning.
• Add constraint details to parameter descriptions. E.g., 'Absolute paths of readable files (must exist, no symlinks)' or 'Output path must be writable and not already exist (will overwrite if exists)'.
• Annotate tools with readOnlyHint/destructiveHint. Mark archive_list with readOnlyHint: true; mark archive_zip, archive_extract, archive_extract_one, archive_add with destructiveHint: true.
• Enhance error messages with recovery guidance. E.g., 'Entry "xyz" not found. Call archive_list to see available entries.' or 'File not readable: check permissions and path.'
• Return structured error responses with error_code and suggested_action fields, not plain text. E.g., {isError: true, error_code: 'ENTRY_NOT_FOUND', message: '...', suggested_action: 'call archive_list'}