MCP Server · Online

Get started

Connect your AI agents to your team via Slack. Choose your client below to get set up in under a minute.

Claude Code (CLI)
Install via slash commands in the terminal
1

Add the Halyard marketplace and install the plugin:

Claude Code
/plugin marketplace add halyard-labs/claude-plugin-marketplace
/plugin install halyard@halyard-labs

Note: If your environment uses sandbox settings that restrict network egress (e.g. Claude Code with Deny networking), you'll need to allow outbound connections to mcp.usehalyard.ai and api.usehalyard.ai for the plugin to work.

2

When prompted, authorize with your Halyard account. The plugin handles OAuth automatically.

3

That's it. Claude will now proactively use ask_expert when it encounters ambiguity, needs requirements clarification, or wants human judgment on a decision.

Claude Desktop
Add via MCP config or plugin/connector

Add the following to your Claude Desktop configuration file:

claude_desktop_config.json
{
  "mcpServers": {
    "halyard": {
      "type": "http",
      "url": "https://mcp.usehalyard.ai"
    }
  }
}

Config file location

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

When first connecting, Claude Desktop will handle the OAuth flow to authorize with your Halyard account.

IDEs: Cursor, Windsurf & others
Add to your editor's MCP config

Add the following to your editor's MCP configuration file:

.cursor/mcp.json / mcp_config.json
{
  "mcpServers": {
    "halyard": {
      "type": "http",
      "url": "https://mcp.usehalyard.ai"
    }
  }
}

Config file locations

Cursor: .cursor/mcp.json in your project root

Windsurf: ~/.codeium/windsurf/mcp_config.json

When first connecting, your editor will handle the OAuth flow to authorize with your Halyard account.

OpenAI Codex & ChatGPT
Connect via remote MCP server
1

In your project's codex.json or through the Codex CLI, add Halyard as a remote MCP server:

codex.json
{
  "mcpServers": {
    "halyard": {
      "type": "http",
      "url": "https://mcp.usehalyard.ai"
    }
  }
}
2

On first use, you'll be redirected to Halyard for OAuth authorization. Sign in with your account to grant access.

3

The available tools will be discovered automatically. You can now ask your agent to consult your teammates for guidance.

Any MCP Client
Generic setup for all MCP-compatible tools

Halyard exposes a standard Model Context Protocol server over HTTP with OAuth 2.0 authentication.

Endpoint

https://mcp.usehalyard.ai POST

Authentication

The server uses OAuth 2.0 with Bearer token authentication. Clients that support RFC 9728 (Protected Resource Metadata) will automatically discover the authorization server via the /.well-known/oauth-protected-resource endpoint.

For manual setup, the authorization server is located at https://api.usehalyard.ai.

Connection config

Point your MCP client at the endpoint URL. The minimum configuration is:

mcp config
{
  "type": "http",
  "url": "https://mcp.usehalyard.ai"
}

Once connected, your agent gains access to these tools. It will use them proactively when it encounters ambiguity or needs human judgment.

list_team
List team members with roles, skills, and expertise profiles
ask_expert
Send a question to the right teammate via Slack
check_response
Wait for and retrieve teammate responses
reply_to_expert
Send follow-up messages in a conversation
get_conversation
Get full conversation history with a teammate
summarize_conversation
Capture teammate answers into the knowledge base
summarize_work
Save implementation decisions and processes
search_knowledge
Search past answers and organizational knowledge
list_knowledge
Browse recent work, decisions, and knowledge by person or time period
list_events
Filter raw work events (PRs, reviews, deploys) by type, repo, author, time
get_delivery_metrics
PR throughput, cycle time, and per-author cadence (admin only)
close_conversation
Close a conversation when done