Skip to Content
FeaturesAI Agent Access (MCP)

AI Agent Access (MCP)

ArchRepo supports the Model Context Protocol (MCP) , allowing you to connect AI agents such as Claude Desktop or Cursor directly to your project’s architecture model.

Once connected, your AI agent can browse model items, read architectural narratives, explore the metamodel, and navigate relationships — all using your own ArchRepo identity.

MCP access is disabled by default. A project admin must enable it in Project Settings before any AI agent can connect.

Prerequisites

Before connecting, make sure you have:

  • An ArchRepo account at archrepo.com 
  • MCP enabled for your project — ask a project admin to turn on MCP Enabled in Project Settings

On first connection, your AI agent will open a browser window asking you to sign in with your ArchRepo account.

Connecting with Claude Desktop, Cursor, Windsurf, or other MCP clients

Use the Streamable HTTP transport URL:

https://archrepo.com/api/mcp

If your client only supports the legacy SSE transport, use:

https://archrepo.com/api/sse

Your MCP client will prompt you to sign in with your ArchRepo credentials when it first connects.

Available Tools

Once connected, your AI agent has access to the following read-only tools:

Discovery

ToolDescription
list_my_projectsReturns all your MCP-enabled projects with their IDs and names. Start here — your agent calls this automatically to discover which projects are available.

Your agent calls list_my_projects first to discover available projects. You can then refer to a project by name — for example, “look at the Platform project” — and the agent will use the correct IDs automatically.

Metamodel

ToolDescription
get_metamodel_typesLists all architecture item types defined in your project’s metamodel, with their descriptions and concerns.
get_metamodel_type_detailsReturns the full specification for a single type, including its properties, allowed relationships, and flags.

Model Items

ToolDescription
list_model_itemsLists items of a given type within your project, with status, readiness score, and key metadata. Supports pagination via limit and offset parameters — the response includes a pagination object with a hasMore flag.
get_model_itemReturns a summary of a model item by ID: core metadata and the full readiness score breakdown. This is a subset of all fields — use get_all_model_item_properties for the complete document.
get_all_model_item_propertiesReturns the complete model item document: all base fields plus type-specific fields (risk scores, assumption status, design decisions, FMEA analysis, API protocols, etc.), relationships, external references, approval state, data mappings, and PlantUML diagrams. Excludes narrative (use get_model_item_narrative) and DataSet entity diagrams (use get_data_set_schema).
get_model_item_narrativeReturns the full architectural narrative for a model item as readable Markdown. This includes design decisions, pipeline descriptions, assumptions, TBDs, and embedded diagrams. See Narratives & Diagrams below.
get_related_model_itemsReturns items connected to a given item, including their description, status, readiness score, and estimation.
search_model_itemsFuzzy full-text search across all model items in a project. Returns up to 30 results ranked by relevance with highlighted match snippets. Searches name, ref, description, external references, relationships, transition states, and more. Optionally filter by ofType. Use this when you don’t know an item’s exact ID.

Data Sets

ToolDescription
get_data_set_schemaReturns the complete entity model for a Data Set: all entities (tables, CSV, JSON, XML, Text, Type definitions) with full attribute definitions including field names, types, constraints, and examples; all relationships between entities with cardinality; and a Mermaid ERD diagram.

Tasks

ToolDescription
fetch_my_tasksReturns the current user’s active tasks (To Do + In Progress) assigned to them across all projects, sorted by progress then due date. No parameters required — identity is resolved from the OAuth token.

Relationships

ToolDescription
get_valid_relationshipsReturns relationship context from the metamodel. With one type: returns full inbound (what points TO this type) and outbound (what this type points TO) maps. With two types: returns candidates in both directions showing which item will own the relationship. Relationship entries may include an optional helpGuidance field with context from the metamodel author on when and how to use that relationship type.
relate_model_itemsCreates a relationship between two model items. The metamodel determines ownership and type automatically. Accepts an optional description to record why the items are related. Returns options when multiple relationship types are valid between the pair.
unrelate_model_itemsRemoves all relationships between two model items, scanning both directions regardless of which item owns the storage.
update_relationship_descriptionSets or updates the description on an existing relationship between two items. Use to add rationale to relationships that were created without one, or to refine existing descriptions.

At session start: call fetch_my_tasks to understand the user’s current workload before diving into analysis.

When you don’t know the item ID: call search_model_items first to locate the item by keyword, then use the returned id with the tools below.

For a thorough analysis of any architecture item, call the tools in this order:

  1. get_model_item — get core metadata and the readiness score, which summarises completeness across all relationship categories
  2. get_all_model_item_properties — get the complete document: external references, relationships, approval state, and all type-specific fields
  3. get_model_item_narrative — read the full architectural description, design decisions, and diagrams
  4. get_related_model_items — explore connected assumptions, risks, requirements, data sets, and operational items
  5. get_data_set_schema — for any related Data Set items, fetch the full entity model and field definitions

All access is scoped to a single project. Your agent cannot read data from projects you are not a member of.

Prompts

In addition to tools, the ArchRepo MCP server provides prompts — pre-built workflow templates that MCP clients can surface as slash commands or quick-start options. Each prompt encodes domain knowledge and the correct tool-calling sequence for a common architectural analysis task.

PromptArgumentsDescription
analyse-architecture-itemorgId, projectId, itemIdentifierFull 5-step analysis of a specific item: metadata, complete properties, narrative, related items, and data models. Produces a structured report with completeness gaps and recommendations.
project-health-checkorgId, projectIdSurvey the entire project for completeness gaps, low-readiness items, open risks, and unresolved assumptions. Produces a RAG-status health report with prioritised recommendations.
impact-assessmentorgId, projectId, itemIdentifier, proposedChangeTrace the ripple effects of a proposed change through related items, data contracts, and risks. Suggests using relate_model_items to add any missing connections identified.
specification-readiness-revieworgId, projectId, itemIdentifierReview whether an item is ready to hand to a development team. Produces a RAG-status checklist of blockers and recommendations. Suggests using relate_model_items to fix missing relationship blockers.
data-lineage-traceorgId, projectId, startingPointTrace data from source through streams, APIs, and services to consuming systems. Identifies transformations and lineage gaps.

In Claude Desktop, prompts appear when you click the + button in the chat input. In other MCP clients they may appear as slash commands.

Narratives & Diagrams

The get_model_item_narrative tool returns the narrative as Markdown, converted from ArchRepo’s rich block editor format. It includes:

  • Headings, paragraphs, lists, and tables exactly as authored
  • Draw.io diagrams as image URLs pointing to ArchRepo’s CDN (https://blobs.archrepo.com/…). AI agents that support image rendering (such as Claude) can view these diagrams directly.
  • PlantUML diagrams as plantuml code blocks. Claude and other AI tools can read PlantUML syntax and understand the diagram’s structure.
  • References to embedded model items and data entities, shown as inline notes with their IDs.

If a model item has no narrative, the tool returns (No narrative content).

Enabling MCP for a Project (Admins)

To allow AI agents to connect to a project:

  1. Go to Project Settings for the project
  2. Toggle MCP Enabled to on
  3. Save the settings

Only project members can connect. Non-members receive an error even if MCP is enabled.

Last updated on