Server has 6 tools with basic schemas and descriptions, but significant gaps limit production readiness. Tool names follow verb_noun convention (generate_image, check_session, list_options). Descriptions exist but are often generic or lack actionable context. Input schemas are present with type definitions, but parameter descriptions are sparse or missing. Output schemas are not documented. Error handling is minimal, no recovery guidance or categorization. The server lacks tool annotations (readOnlyHint, destructiveHint, idempotentHint) and does not guide LLMs on which tools are safe to retry. Composition is reasonable (separate image/video tools), but parameter naming could be more explicit (e.g., 'model' is ambiguous without enum constraints).

• Add explicit output schemas for all tools. Document return fields (file_path, image_content, session_state, available_models, etc.) with types and descriptions so LLMs can chain calls.
• Convert 'model' parameter to enum with explicit list of supported models (GPT Image 2, Flux 2, Recraft V3, etc.) instead of relying on partial-match logic. Add description: 'AI model to use. Supports partial name matching (e.g., "Flux" matches "Flux 2").'
• Add tool annotations: mark generate_image and generate_video with destructiveHint=true (they write files); mark check_session and check_video_session with readOnlyHint=true.
• Enhance error descriptions with recovery paths. E.g., 'If session is invalid (not_logged_in), run login.py in headful mode to re-authenticate, then retry.'
• Standardize parameter naming and values. Replace mixed-language tier values with English enums (lite_1x, standard_2_5x, ultra_5x). Add min/max constraints for numeric params (duration: 1 - 120 seconds, timeout: 30 - 3600 seconds).