invoking-tools-and-prompts/understanding-responses
Understanding Responses
How MCPFlo renders content blocks — JSON tree, images, audio, embedded resources, and mixed content.
Every tool call, resource read, or prompt render in MCPFlo returns one or more content blocks, and MCPFlo renders each block according to its type — so you see the response the way it’s meant to be read, not just as raw text.
Content block types
Text Rendered as plain, readable text.
JSON Pretty-printed with a collapsible tree view — expand/collapse nested objects and arrays to navigate large payloads without scrolling through a wall of text.
Images Rendered inline as actual images, not base64 strings.
Audio Rendered with an inline player.
Embedded / linked resources Resources referenced from within a response (see Reading Resources & Rendering Prompts) render inline using the same type-aware rendering as a direct resource read.
Mixed content
A single response can contain multiple content blocks of different types — for example, a tool that returns an explanatory text block followed by an image. MCPFlo renders mixed content inline, in the order the server returned it, rather than forcing you to pick one view.
Raw view
However a response is rendered, the full JSON-RPC envelope is always available as a raw view — useful when you need to see exactly what the server sent at the protocol level (headers, IDs, error objects, etc.), independent of how MCPFlo chose to display it.
Why this matters
Rich, type-aware rendering means you can verify a server’s output is correct at a glance — spotting a malformed image, garbled JSON, or missing field — without needing to decode or format the raw response yourself.
Related
- See Per-Call Token Footprint for how a response’s size translates into context-window cost.
- See Live Notifications for output that arrives during a call rather than as the final response.