Available MCP Tools

Structured architecture information, available on demand.

SPECTRALE exposes nine read-only MCP tools.

Each tool provides a specific view of the architecture context.

Tool availability depends on the connection's permissions.

Project tools

list_projects

Lists projects accessible to the authorized connection.

Use it when an agent needs to identify which project contexts are available.

get_project_overview

Retrieves an overview of an authorized project.

Use it to establish high-level context before exploring individual components.

Architecture tools

list_components

Lists architectural components identified within a project.

Useful for understanding the main building blocks of an application.

get_component_details

Retrieves additional information about a specific architectural component.

Useful when an agent needs more context about a service, database, repository, or other modeled resource.

get_dependencies

Retrieves dependency relationships between architectural components.

Useful for exploring how changes to one component may relate to other parts of the system.

Discovery tools

get_discovery_status

Retrieves information about the project's discovery state.

Useful for understanding whether discovery results are available and interpreting the freshness or completeness of the available context.

list_boundaries

Lists identified external architectural boundaries.

Useful for identifying systems and services that sit outside the immediate project environment.

Integration metadata tools

list_integrations

Retrieves available integration-related metadata within the authorized project context.

This tool does not provide unrestricted access to external platforms or expose provider credentials.

get_integration_capabilities

Retrieves information about recognized integration capabilities.

Capability metadata should not be interpreted as proof that a third-party connection is active or available for use.

Example workflow

A developer asks an AI agent:

"Help me understand the dependencies of our backend service."

A compatible agent could use the following sequence:

list_projects
    |
    v
get_project_overview
    |
    v
list_components
    |
    v
get_component_details
    |
    v
get_dependencies
    |
    v
ARCHITECTURAL CONTEXT

The agent can then use the retrieved information to inform its response.

The exact sequence depends on the AI client and the user's request.

Important limitations

MCP tools return information available in SPECTRALE's architecture model.

They do not guarantee that every runtime dependency or external system has been discovered.

Results should be interpreted in the context of the available Discovery evidence.