Clarification Needed on Gamma Claude Connector IDs, APIs, Rendering Status, Image Options, and Usage Limits

Hi — we use the Gamma Claude Connector (MCP: generate, read_gamma, get_themes, get_folders) to generate presentations programmatically. We've hit several ambiguities between the connector, REST API, and public MCP docs. Could you clarify the following? IDs and URLs Is generationId ever the same as the final gammaId, or always distinct?

Polling and status 4. Does the Claude Connector's generate tool block until render completes and return a final URL, or return immediately with pending status? 5. If it returns pending, how do we check status via the connector? We don't see a status tool in the schema. 6. Is GET /v1.0/generations/{id} (REST) exposed through the connector in any form? 7. What's the recommended pattern to confirm "render complete" from a connector client without opening a browser? imageOptions.source enum drift 8. Three sources disagree — REST API guide lists themeAccent; MCP tools reference lists unsplash; live connector schema has neither. Which is canonical? 9. How do we pass themeAccent via the Claude Connector? It rejects the value. 10. Status of unsplash — renamed, deprecated, still valid somewhere? Real image rendering 11. Your image URL guide says markdown  is skipped by the preprocessor. True under all imageOptions.source values, or only some? 12. Under noImages: do raw HTTPS URLs and markdown URLs both render, or only raw URLs? 13. Under aiGenerated (or any non-noImages value): if we provide raw URLs in inputText, are they rendered alongside AI images, or replaced? 14. Is there a setting explicitly meaning "use only my provided URLs — do not generate, source, or substitute anything"? Multi-image slide layout 15. When a slide has three images of the same aspect ratio (e.g. all 9:16 vertical), does Gamma preserve the ratio or force a uniform crop? 16. Can we hint gallery layout (grid / row / portrait strip) via additionalInstructions, or is there a structured parameter? Connector-specific 17. read_gamma is in our live connector but not on the public MCP tools reference page — connector-only, or undocumented? 18. Is there a canonical schema doc for the connector specifically? (We currently infer from the live tool descriptions.) 19. Is there an Anthropic-maintained or Gamma-maintained Claude skill / reference implementation for Gamma? Limits and reliability 20. Confirmed inputText character limit? 21. Published rate limits (burst / hourly / daily)? 22. On large payloads (~21 slides with ~12 image URLs), we've seen the generate tool timeout or return before rendering completes. Is there a silent timeout? Our goal: avoid believing a generation has failed when it has actually succeeded, avoid themeAccent failures via the connector, avoid markdown-syntax image URLs being silently dropped, and avoid any silent AI replacement of our real image URLs. Links to canonical references alongside answers would really help. Thanks.