Overview

Sequa.AI MCP Server exposes Sequa’s contextual knowledge engine over the Model Context Protocol (MCP). It lets developer tools and AI assistants connect to a live view of your codebase, docs, tickets, and internal standards so generated documentation and assistant responses stay aligned with the actual repository state. Instead of depending on manually-maintained pages that drift out of date, Sequa continuously derives accurate docs (architecture summaries, API surface, runbooks, onboarding guides) from your source.

The project in this repository is the MCP server entrypoint for Sequa. Install it as an MCP server in any MCP-capable client (Cursor, Claude, VS Code, Windsurf, etc.) to give agents real-time access to project-specific context via the MCP transport chosen in your Sequa project. A working Sequa project and the MCP setup URL are required before accepting connections.

Features

Continuous extraction of up-to-date documentation from source code and repo metadata
MCP server that integrates with any MCP-aware client or agent
Project-scoped authentication: only the Sequa project you configure may connect
Support for common local-launch workflows (npx proxy) so no global install is necessary
Usable from IDEs, desktop clients, CLI tools, and hosted agent UIs
Links back to source (files, commits, PRs) in generated answers

Installation / Configuration

Prerequisites:

Create or sign in to a Sequa account at https://app.sequa.ai/login
Create a Project inside Sequa
In the project, copy the MCP Setup URL for the transport your client supports

Quick start (launch a local proxy with npx):

npx -y @sequa-ai/sequa-mcp@latest https://mcp.sequa.ai/v1/setup-code-assistant

Replace the example URL with the project-specific MCP URL from the Sequa app.

Example client configurations

Cursor (~/.cursor/mcp.json)

{
  "mcpServers": {
    "sequa": {
      "url": "https://mcp.sequa.ai/v1/setup-code-assistant"
    }
  }
}

Claude Desktop (edit Developer → Config)

{
  "mcpServers": {
    "sequa": {
      "command": "npx",
      "args": [
        "-y",
        "@sequa-ai/sequa-mcp@latest",
        "https://mcp.sequa.ai/v1/setup-code-assistant"
      ]
    }
  }
}

VS Code (.vscode/mcp.json)

{
  "servers": {
    "sequa": {
      "command": "npx",
      "args": [
        "-y",
        "@sequa-ai/sequa-mcp@latest",
        "https://mcp.sequa.ai/v1/setup-code-assistant"
      ]
    }
  }
}

CLI / tools that accept a command (e.g., Highlight, Augment)

npx -y @sequa-ai/sequa-mcp@latest https://mcp.sequa.ai/v1/setup-code-assistant

Important: if no Sequa project is configured or the MCP URL is invalid, the proxy will refuse connections and return auth/project errors.

Available Tools / Resources

Resource	Purpose
GitHub repo	https://github.com/sequa-ai/sequa-mcp — source, issues, and contribution notes
Sequa app	https://app.sequa.ai/login — create projects and copy MCP setup URLs
MCP spec	Use your client’s MCP documentation for server registration details

The repo contains the MCP server entrypoint and examples for multiple clients; use the project-specific MCP URL from the Sequa app when launching the proxy.

Use Cases

Onboarding new engineers
- Generate an up-to-date architecture overview and a “where to find things” guide derived from your repos and monorepo layout. New hires get accurate links and file references instead of stale Confluence pages.
AI-powered code assistance
- When an LLM-powered assistant uses Sequa via MCP, it can answer questions like “How should I call the billing API?” or “Which service owns user notifications?” with current code references and URLs to the exact source lines.
Code review and PR context
- Attach live context to code review tools or editors so diffs reference the most recent API docs and runbook steps; reviewers save time by not hunting for scattered notes.
Automated runbooks and incident response
- Keep runbooks in sync with code and deployment manifests; when on-call engineers query “how to rotate keys for service X,” the guidance will reflect the present repo and infra config.
Consolidating scattered knowledge
- Combine README fragments, issue tracker notes, and internal standards into a single contextual layer available to humans and agents.

Troubleshooting and Tips

Ensure you created a Sequa project and copied the exact MCP URL; missing projects are the most common cause of connection failures.
When testing locally, use the npx command with the project URL to quickly verify the client integration before adding permanent configuration files.
For offline development or limited network environments, confirm your client supports launching MCP command-style servers (many desktop and IDE integrations do).

For code, bug reports, or pull requests, see the GitHub repository: https://github.com/sequa-ai/sequa-mcp.

Sequa.AI MCP Server: Self-Maintaining Code Docs