doesitsupport

← all clients

Cursor CLI

Category: cli · Last checked: 2026-07-04 · Tested version: 2026.07.01-41b2de7

Capability support

FeatureStatusNotes
Image renderingunsupported2026-07-04T14:44:07.313ZTool was called and the model echoed the probe's call-id back to the user — but a terminal/code-exec surface can't render images or iframes. Default unsupported.
Animated GIFunsupported2026-07-04T14:44:18.656ZTool was called and the model echoed the probe's call-id back to the user — but a terminal/code-exec surface can't render images or iframes. Default unsupported.
SVGunsupported2026-07-04T14:44:29.741ZTool was called but the model did not echo the probe's call-id. Could be a render (no quote needed) or a swallow. Default unsupported for tier.
HEICunsupported2026-07-04T14:44:34.992ZTool was called but the model did not echo the probe's call-id. Could be a render (no quote needed) or a swallow. Default unsupported for tier.
MCP Apps HTML widgetunsupported2026-07-04T14:44:40.864ZTool was called and the model echoed the probe's call-id back to the user — but a terminal/code-exec surface can't render images or iframes. Default unsupported.
ChatGPT Apps SDK componentunsupported2026-07-04T14:44:45.316ZTool was called and the model echoed the probe's call-id back to the user — but a terminal/code-exec surface can't render images or iframes. Default unsupported.
Markdown: basicsupported2026-07-04T14:44:50.766ZMarkdown content was returned and the model echoed it (including the call-id) back to the user.
Markdown: tablespartial2026-07-04T14:44:56.828ZTool was called and the model echoed the probe's call-id — but behavioral can't tell whether the client actually rendered the markdown variant (table/code/math/diagram/image/task/URL) vs. surfaced the source. Prefer vision-tier for a definitive answer.
Markdown: code blockspartial2026-07-04T14:45:01.410ZTool was called and the model echoed the probe's call-id — but behavioral can't tell whether the client actually rendered the markdown variant (table/code/math/diagram/image/task/URL) vs. surfaced the source. Prefer vision-tier for a definitive answer.
Markdown: imagespartial2026-07-04T14:45:06.872ZTool was called and the model echoed the probe's call-id — but behavioral can't tell whether the client actually rendered the markdown variant (table/code/math/diagram/image/task/URL) vs. surfaced the source. Prefer vision-tier for a definitive answer.
Markdown: task listspartial2026-07-04T14:45:13.990ZTool was called and the model echoed the probe's call-id — but behavioral can't tell whether the client actually rendered the markdown variant (table/code/math/diagram/image/task/URL) vs. surfaced the source. Prefer vision-tier for a definitive answer.
URL autolinkingpartial2026-07-04T14:45:19.312ZTool was called and the model echoed the probe's call-id — but behavioral can't tell whether the client actually rendered the markdown variant (table/code/math/diagram/image/task/URL) vs. surfaced the source. Prefer vision-tier for a definitive answer.
MCP resourcessupported2026-07-04T15:43:15.891ZThe model's reply quoted the render token that exists only in the resource content block's body — the client surfaced the resource.
MCP prompt templatesunknown2026-07-04T14:45:29.944ZMCP protocol feature; behavioral classifier can't determine rendering — defer to vision or manual review.
MCP elicitationunsupported2026-07-04T15:42:53.445ZClient did not declare the elicitation capability during MCP initialize.
MCP samplingunsupported2026-07-04T15:43:00.496ZClient did not declare the sampling capability during MCP initialize.
MCP structured contentunsupported2026-07-04T15:46:08.084ZThe model echoed the tool response's nonce but referenced none of the structuredContent payload fields — the client likely surfaced only the text fallback.