Tools aren’t the only capability you can invoke in MCPFlo — resources and prompts have their own interaction patterns, distinct from a tool call.

Reading resources

A resource is data the server can hand over, identified by a URI — reading one has no side effects, it’s a pure fetch.

1. Select a resource in the capability tree.
2. MCPFlo requests its contents from the server.
3. The result renders based on content type:
   • Text resources render as readable text (with syntax highlighting where applicable).
   • Binary resources (images, audio, etc.) render using the same rich rendering as tool responses — see Understanding Responses.
4. The raw JSON-RPC response is always available if you need to inspect it directly.

Resources can also appear embedded or linked inside a tool’s response — in that case, MCPFlo renders them inline as part of the result, without you needing to navigate to the resource separately.

Rendering prompts

A prompt is a reusable, parameterized template for a conversation — MCPFlo lets you expand one to see exactly what message turns it would produce.

1. Select a prompt in the capability tree.
2. If the prompt declares named arguments, a form renders for you to fill them in (schema-driven, same as tool inputs).
3. Submit the form — MCPFlo requests the rendered prompt from the server.
4. The response shows the expanded message turns (e.g. system/user/assistant messages) exactly as the server constructed them, with each turn’s role and content clearly labeled.

This is useful for verifying a prompt template behaves correctly — that arguments are substituted properly, roles are assigned as expected, and the resulting conversation shape matches what you intended before wiring it into a real client.

Related

• See Understanding Responses for how different content types (text, JSON, images, audio) are rendered once a resource or prompt call returns.
• See Per-Call Token Footprint for how much context budget a given resource read or prompt render consumes.