Connecting AI Tools

Integrate Agiflow into your existing AI workflow via Model Context Protocol (MCP). Enable your agents to query project state, manage tasks, and follow architecture rules.

01 // Quick Setup

Add the following configuration to your MCP settings file. Replace {orgId} with your organization identifier.

MCP_CONFIG_DEFAULT

{
  "mcpServers": {
    "agiflow": {
      "url": "https://api.agiflow.io/api/v1/organizations/{orgId}/mcp"
    }
  }
}

💡 Session Authorization:

On first tool execution, Agiflow will trigger an OAuth browser flow. Grant access to your organization to establish a persistent session.

Secure Transport

All tool calls are encrypted and scoped to your organization boundaries.

02 // Scope Matrix

Scope_Level

Endpoint_Pattern

Context_Boundary

Organization

/organizations/{orgId}/mcp

Full management across all projects

Project

/projects/{projectId}/mcp

Isolation to a specific product stack

Work Unit

/work-units/{workUnitId}/mcp

Restricted to a single feature lifecycle

Task

/tasks/{taskId}/mcp

Atomic access to a single work item

03 // Authentication

OAuth_Standard

Recommended for interactive tools like Claude Code and Cursor. Seamlessly handles token rotation and organization selection via browser flow.

Multi-org support
Zero configuration setup
Persistent sessions

Static_Token

Best for CI/CD pipelines, headless agents, or automated scripts. Generate long-lived keys from Dashboard → Settings → API Keys.

AUTH_HEADER_SNIPPET

"headers": {
  "Authorization": "Bearer {key}"
}

04 // Discovery

describe_capabilities

Recursive discovery of tools and domain skills

use_capability

Dynamic execution of any discovered capability

get_current_scope

Runtime verification of session context

Ready to integrate?

Initialize your environment by connecting your primary AI tool. No credit card required for standard orchestration.

INIT_AUTH_FLOW VIEW_QUICKSTART