After Effects MCP has 13 tools with complete input schemas and descriptions, but quality is uneven. Tool names follow verb_noun convention (ae_exec, ae_read, ae_checkpoint) which is good. However, descriptions are often generic or lack actionable context for LLM selection. Parameters are typed but many lack detailed constraints or format guidance. Output schemas are not documented in the provided source. Error handling is minimal, no recovery guidance or categorization. The tool library integration (ae_toolSearch, ae_toolUse, ae_toolSave) is sophisticated but underdocumented. STDIO transport caps protocol readiness at 50.

• Document output schemas for all tools. For ae_exec, specify: {success: boolean, result?: any, error?: string}. For ae_previewFrame: {data: base64_string, width: number, height: number}. For ae_read: {data: object, query: string}.
• Enhance ae_read and ae_nativeExec descriptions with format examples. E.g. ae_read: 'Query syntax: layer.name, comp.duration, project.items[0].name. Returns matching property values as JSON.'
• Add error recovery guidance to WRITE tools. E.g. ae_exec: 'If execution fails, call ae_execRecover with the code and error message to diagnose the issue.'
• Clarify tool library workflow in descriptions. ae_toolSearch: 'Search for saved tools before writing new code. Returns artifact IDs.' ae_toolUse: 'Execute a saved tool by ID from ae_toolSearch results.'
• Add numeric constraints to ae_exec timeout: 'Timeout in milliseconds (100 - 300000). Defaults to 5000.' Similarly constrain ae_previewFrame width/height (e.g. 64 - 4096 pixels).