LiteCLI is a CLI compiler framework, not a production MCP server. Descriptions vary widely: some are detailed (coding_agent: 287 chars), others are minimal (canvas, notion, openai_image_gen, openai_whisper, openai_whisper_api, trello, video_frames: all {}). Tool names lack consistent verb-noun patterns (canvas, notion, trello are nouns; coding_agent is compound). No parameter descriptions visible for empty-schema tools. No output schemas documented. Error handling guidance absent. The 'add' tool has a well-formed schema with 10 parameters and descriptions, but most others are stubs or wrappers around external APIs without proper MCP-level schema definition.

• Add explicit input schemas to all 9 stub tools (canvas, notion, openai_image_gen, openai_whisper, openai_whisper_api, trello, video_frames, weather). Even if a tool takes no parameters, declare 'input: {type: object, properties: {}, required: []}' rather than empty {}.
• Expand descriptions for stub tools to 50 - 200 characters. Example: 'canvas' → 'Create and manage collaborative canvas sketches for visual brainstorming and design collaboration.' Include WHEN to use and WHAT it returns.
• Document output schemas for all tools. Specify returned fields, types, and which IDs/references downstream tools need. Example: 'Returns {canvas_id, title, created_at, collaborators: [{user_id, email}]}'.
• Rename tools to follow verb_noun pattern: 'canvas' → 'create_canvas', 'notion' → 'create_notion_page', 'trello' → 'create_trello_card', 'weather' → 'get_weather'. This helps LLMs infer intent from names alone.
• Add error handling guidance to all tools. Example: 'If API returns 401, check credentials. If 429, retry after 60s. If 404, call discover_backends() to list available services.' Include recovery paths for common failures.