HAOps MCP Server

HAOps MCP Server

Enables Claude to interact with the HAOps project management system for managing projects, modules, features, and issues.

Category
Visit Server

README

HAOps MCP Server

Model Context Protocol (MCP) server for HAOps (Human-Agent Operations) - enables Claude to interact with HAOps project management system.

Features

  • Resources: Access projects, modules, features, and issues
  • Tools: Create, update, and manage HAOps entities via Claude
  • API Key Authentication: Secure access using HAOps API keys

Setup

Prerequisites

  • Node.js >= 18.0.0
  • HAOps instance running (local or remote)
  • HAOps API key (generate via Admin → API Keys)

Installation

cd haops-mcp-server
npm install

Configuration

Create .env file:

HAOPS_API_URL=http://localhost:3000
HAOPS_API_KEY=your-api-key-here

Build

npm run build

Development

npm run dev

Usage

Claude Code (CLI / VS Code Extension)

Register the MCP server via the Claude CLI:

claude mcp add-json --scope user haops '{
  "type": "stdio",
  "command": "node",
  "args": ["/path/to/feature-tracker/haops-mcp-server/dist/index.js"],
  "env": {
    "HAOPS_API_URL": "http://localhost:3000",
    "HAOPS_API_KEY": "your-api-key-here"
  }
}'

Verify: claude mcp list should show haops: ... - ✓ Connected

Note: Do NOT manually edit ~/.claude/mcp-servers.json — Claude Code does not read that file. Always use claude mcp add-json.

Claude Desktop (App)

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "haops": {
      "command": "node",
      "args": ["/path/to/feature-tracker/haops-mcp-server/dist/index.js"],
      "env": {
        "HAOPS_API_URL": "http://localhost:3000",
        "HAOPS_API_KEY": "your-api-key-here"
      }
    }
  }
}

Manual Testing

Run the server manually:

npm start

The server will listen on stdio for MCP protocol messages.

HTTP Mode (Shared Daemon)

In addition to the default stdio transport, the server can run as a long-lived HTTP daemon. One daemon process can serve multiple concurrent Claude clients (CLI, VS Code extension, desktop app). Because every session shares the same HAOps API client singleton, this saves hundreds of MB of RAM compared to launching one stdio subprocess per Claude session.

Starting the daemon

# Default port (3100)
HAOPS_API_URL=http://localhost:3000 HAOPS_API_KEY=your-key \
  node dist/index.js --http

# Custom port
HAOPS_API_URL=http://localhost:3000 HAOPS_API_KEY=your-key \
  node dist/index.js --http --port 3199

The daemon binds 127.0.0.1 only and validates the Host header on every request (DNS rebinding protection). It is not suitable for exposure beyond localhost — treat it as a developer-machine convenience.

On SIGTERM / SIGINT the daemon closes all live MCP sessions cleanly and exits 0.

Endpoints

Method Path Purpose
POST /mcp MCP JSON-RPC (initialize, tools/list, tools/call, ...)
GET /mcp SSE stream for server-initiated notifications (per session)
DELETE /mcp Explicit session close
GET /health Liveness probe — { status, uptime, version, connections, sessions }

Claude Code client config

# Add the HTTP transport to Claude Code
claude mcp add-json --scope user haops '{
  "type": "http",
  "url": "http://127.0.0.1:3100/mcp"
}'

# Or for a non-default port
claude mcp add-json --scope user haops '{
  "type": "http",
  "url": "http://127.0.0.1:3199/mcp"
}'

Stdio mode remains the default when neither --http nor --port is passed, so existing stdio-based configurations keep working unchanged.

Smoke test with curl

# Health
curl http://127.0.0.1:3100/health

# Initialize a session (save the Mcp-Session-Id header from the response)
curl -i -X POST http://127.0.0.1:3100/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -d '{"jsonrpc":"2.0","method":"initialize","params":{"protocolVersion":"2025-03-26","capabilities":{},"clientInfo":{"name":"curl","version":"0"}},"id":1}'

# List tools (reuse the session id from above)
curl -X POST http://127.0.0.1:3100/mcp \
  -H "Content-Type: application/json" \
  -H "Accept: application/json, text/event-stream" \
  -H "Mcp-Session-Id: <session-id-from-init>" \
  -H "Mcp-Protocol-Version: 2025-03-26" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":2}'

When to use which mode

Scenario Transport
Single Claude Code session, default setup stdio
Multiple Claude Code sessions on the same machine HTTP daemon
Claude Desktop stdio
VS Code extension (alongside CLI sessions) HTTP daemon

Architecture

haops-mcp-server/
├── src/
│   ├── index.ts          # MCP server entry point
│   ├── api/
│   │   └── client.ts     # HAOps API client (HTTP)
│   ├── resources/        # MCP resources (projects, modules, etc.)
│   └── tools/            # MCP tools (CRUD operations)
├── dist/                 # Compiled JavaScript (generated)
└── package.json

Resources

MCP resources exposed by this server (with query filtering):

  • haops://projects - List all projects
  • haops://projects/{slug}/modules?status=&priority=&ownerId= - List modules
  • haops://projects/{slug}/features?status=&priority= - List features
  • haops://projects/{slug}/issues?status=&priority=&assignedTo=&type= - List issues

Quiet Mode (v2.8)

All write tools (create/update/delete/claim/post/etc.) return a compact one-line summary by default to minimize context token usage:

Created — abc-123 [in-progress] "My new issue"

To get the full API response, pass verbose: true to any write tool:

{ "issueId": "...", "title": "...", "verbose": true }

The special haops_append_memory tool returns {id, timestamp, tag} in compact mode (no echo of the content you just wrote). Read tools are unaffected.

Telemetry (v2.8)

The server records response byte counts and call counts per tool to ~/.haops-mcp/stats/YYYY-MM-DD.jsonl (fire-and-forget, zero hot-path overhead).

View a report:

npm run stats             # today, top 20 tools by bytes
npm run stats:week        # last 7 days combined
npx tsx scripts/mcp-stats.ts --days 30 --top 10

Example output:

MCP Telemetry Report — today (2026-06-11)
Total calls: 147 | Total bytes: 312.4 KB

Tool                                                Calls  Total bytes  Avg bytes
------------------------------------------------------------------------------------
haops_read_memory                                      23      142.3 KB      6.3 KB
haops_read_protocol                                    18       89.1 KB      4.9 KB
haops_list_issues                                      31       44.2 KB      1.4 KB
haops_create_issue                                     12          743 B         62 B

Tools

MCP tools provided by this server (115 total). Key tools listed below — see src/index.ts for full list:

Read Operations

  • haops_list_projects() - List all projects
  • haops_list_members(projectSlug) - List project members with roles and stats

Module CRUD

  • haops_create_module(projectSlug, moduleData) - Create new module
  • haops_update_module(projectSlug, moduleId, moduleData) - Update module
  • haops_delete_module(projectSlug, moduleId, confirm?) - Delete module (cascade safety check)

Feature CRUD

  • haops_create_feature(projectSlug, featureData) - Create new feature
  • haops_update_feature(projectSlug, featureId, featureData) - Update feature
  • haops_delete_feature(projectSlug, featureId, confirm?) - Delete feature (cascade safety check)

Issue CRUD

  • haops_create_issue(projectSlug, issueData) - Create new issue
  • haops_update_issue(projectSlug, issueId, issueData) - Update issue
  • haops_delete_issue(projectSlug, issueId) - Delete issue
  • haops_bulk_update_issues(projectSlug, issueIds, updates) - Bulk update issues

Communication

  • haops_create_discussion(projectSlug, title, ...) - Create discussion thread
  • haops_post_message(projectSlug, discussionId, content) - Post to discussion
  • haops_send_dm(projectSlug, recipientUserId, content) - Send direct message

Team Management

  • haops_add_member(projectSlug, userId, role?) - Add project member
  • haops_update_member_role(projectSlug, userId, role) - Update member role

Audit & Activity

  • haops_get_activity(projectSlug, entityType, entityId) - Entity activity log
  • haops_get_audit_log(page?, limit?, action?, entityType?) - System audit log (admin)

Tool Usage Examples

1. Create a Module

{
  "projectSlug": "my-project",
  "title": "Authentication Module",
  "description": "User authentication and authorization",
  "ownerId": "user-uuid-here",
  "status": "in-progress",
  "priority": "high",
  "startDate": "2026-02-24",
  "targetDate": "2026-03-15"
}

Required fields: projectSlug, title, ownerId

Status options: backlog, in-progress, review, done, blocked, on-hold, cancelled

Priority options: low, medium, high, critical

2. Update a Module

{
  "projectSlug": "my-project",
  "moduleId": "module-uuid-here",
  "status": "done",
  "completedDate": "2026-03-10"
}

Required fields: projectSlug, moduleId

All other fields are optional - only include fields you want to change.

3. Create a Feature

{
  "projectSlug": "my-project",
  "moduleId": "module-uuid-here",
  "title": "OAuth2 Integration",
  "description": "Support Google and GitHub OAuth",
  "acceptanceCriteria": "Users can sign in with Google or GitHub accounts",
  "ownerId": "user-uuid-here",
  "status": "backlog",
  "priority": "medium"
}

Required fields: projectSlug, moduleId, title, ownerId

Note: Feature has acceptanceCriteria field (optional).

4. Update a Feature

{
  "projectSlug": "my-project",
  "featureId": "feature-uuid-here",
  "status": "review",
  "ownerId": "new-owner-uuid"
}

Required fields: projectSlug, featureId

5. Create an Issue

{
  "projectSlug": "my-project",
  "featureId": "feature-uuid-here",
  "title": "Fix OAuth redirect URL bug",
  "description": "Redirect URL is incorrect after GitHub login",
  "acceptanceCriteria": "Redirect URL points to dashboard after login",
  "type": "bug",
  "status": "backlog",
  "priority": "critical",
  "targetDate": "2026-02-28",
  "assignedTo": "user-uuid-here"
}

Required fields: projectSlug, featureId, title

Type options: feature, bug, task, optimization, refactor, documentation, research

Note: Issue uses assignedTo field (not ownerId like Module/Feature).

6. Update an Issue

{
  "projectSlug": "my-project",
  "issueId": "issue-uuid-here",
  "status": "done",
  "type": "feature",
  "completedDate": "2026-02-25"
}

Required fields: projectSlug, issueId

7. Delete a Module (with safety check)

{
  "projectSlug": "my-project",
  "moduleId": "module-uuid-here"
}

Without confirm: true, returns a warning listing child features/issues. Add "confirm": true to proceed with cascade deletion.

8. Bulk Update Issues

{
  "projectSlug": "my-project",
  "issueIds": ["issue-1-uuid", "issue-2-uuid", "issue-3-uuid"],
  "updates": {
    "status": "done",
    "priority": "high"
  }
}

Updates all specified issues atomically (uses DB transaction with rollback).

9. Create a Discussion

{
  "projectSlug": "my-project",
  "title": "Architecture Review",
  "type": "question",
  "channelId": "channel-uuid-here",
  "firstMessage": "Let's discuss the new auth flow."
}

Supports channel-based, entity-linked (discussableType + discussableId), or both.

10. Send a Direct Message

{
  "projectSlug": "my-project",
  "recipientUserId": "user-uuid-here",
  "content": "Hey, can you review my PR?"
}

Testing

Run Unit Tests

npm test                # All tests
npm run test:unit       # Unit tests only (src/)

Unit tests (22 total):

  • src/api/__tests__/client.test.ts - API client methods (6 tests)
  • Backend: lib/utils/__tests__/apiKeys.test.ts - Key generation/hashing (9 tests)
  • Backend: lib/auth/__tests__/requireApiKey.test.ts - Auth middleware (7 tests)

Run Integration Tests

npm run test:integration    # Integration tests (tests/integration/)
npm run test:all            # All tests (unit + integration)

Integration tests (13 skeleton tests):

  • See tests/integration/README.md for setup instructions
  • Tests are currently templates with commented-out API calls
  • Require test database and seed data to run E2E

Development

Project Structure

haops-mcp-server/
├── src/
│   ├── index.ts              # Main MCP server (stdio transport)
│   ├── api/
│   │   ├── client.ts         # HAOps API HTTP client
│   │   └── __tests__/        # API client unit tests
│   └── types/
│       └── entities.ts       # TypeScript interfaces
├── tests/
│   └── integration/          # Integration tests (E2E)
├── dist/                     # Compiled JavaScript (npm run build)
├── package.json
└── README.md

Adding New Tools

  1. Add TypeScript types in src/types/entities.ts (if needed)
  2. Add API client method in src/api/client.ts
  3. Define tool schema in src/index.ts (ListToolsRequestSchema handler)
    • Include inputSchema with all parameters + required fields
  4. Implement tool logic in CallToolRequestSchema handler
    • Extract args with proper TypeScript typing
    • Build request payload (only include defined fields)
    • Call API client method
    • Return formatted response
  5. Write unit tests in src/api/__tests__/
  6. Add usage example to this README

Code Quality

  • TypeScript: Strict mode enabled, proper typing required
  • Tests: Jest with ts-jest for ESM modules
  • Linting: ESLint with TypeScript plugin
  • Build: npm run build compiles to dist/ directory

Troubleshooting

"HAOPS_API_KEY environment variable is required"

  • Ensure .env file exists in haops-mcp-server/ directory
  • Or set environment variables in Claude Desktop config

"Cannot connect to HAOps API"

  • Check HAOPS_API_URL points to running HAOps instance
  • Verify HAOps server is accessible (try curl $HAOPS_API_URL/api/projects)

"API key authentication failed"

  • Generate new API key at HAOps → Admin → API Keys
  • Copy full key (shown only once)
  • Ensure key is not expired

License

MIT

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured