MCP can represent operations, readable context and reusable prompt templates. What you see depends on the capabilities exposed by your assigned server and supported by your AI application.
Tools#
Tools perform defined operations. A tool has a name and input schema. The AI client can discover the available tools and request a call using those inputs. Your managed connection is verified around this tool-based workflow.
Resources#
A resource is readable context identified by a URI. It can help an application access information without presenting it as a general operation. Do not assume resources are enabled: a tools-only credential can intentionally deny resource access.
Prompts#
A prompt template provides a reusable starting instruction. It does not grant new permissions or prove that a workflow has run. Review its variables and account context before use.
If an option is missing#
Some clients show tools prominently while exposing resources or prompts differently. Others may not support a particular capability. Ask DigLab whether the capability is included in your handoff before treating its absence as an error.
Template and example boundaries#
Examples help explain a workflow but are not a complete tool catalog for your account. Installing or requesting a prepared collection of tool definitions still requires correct service authentication and testing. Use request a custom integration when the work you need is not yet available.
For protocol background, see the MCP architecture overview. The capabilities of your actual connection take precedence over a general protocol feature list.