Server has 7 tools with complete schemas and descriptions. Naming follows verb_noun convention (survey_info, list_objects, import_segy, export_horizon_ascii, autotrack_horizon, read_horizon, compare_horizons). Descriptions are present and substantive (avg ~150 chars), explaining WHAT and WHEN. However, parameter descriptions lack detail on constraints, ranges, and error recovery. Output schemas are not formally documented. Error handling is minimal, no guidance on retryability or recovery paths. Tools are well-composed (single responsibility) and include dry_run patterns for safety. Missing: per-parameter type validation hints, output schema documentation, actionable error messages.

• Document output schemas for all tools. E.g., survey_info returns {ok: bool, install_path: str, survey_name: str, ready: bool}. Use JSON Schema or a structured format in docstrings.
• Add enum constraints to parameters. E.g., event in autotrack_horizon should be enum: ['Max', 'Min', 'Zero+-', 'Zero-+']. Add translator_group examples or enum to list_objects.
• Enhance parameter descriptions with ranges and validation rules. E.g., 'correlation_threshold (0.0 - 1.0, default 0.8): Correlation threshold for tracking' and 'search_window_ms: [start, stop] in milliseconds, e.g. [-20, 20]'.
• Add error handling guidance. Return structured errors with actionable messages: {ok: false, error: 'OpendTect executable not found', recovery: 'Set OPENDTECT_BIN environment variable and restart'}.
• Document dry_run behavior in parameter descriptions, not just return values. E.g., 'dry_run (bool, default True): If true, return the .odcmd script without executing. Review the script against your OpendTect version before setting False.'