Open and Async MCP Server – AI Tools for Async-First Teams

TL;DR

TL;DR: An open-source MCP server that turns async-first working practices into AI tools — draft decision docs, score status updates, run async standups, and get role-aware guidance without leaving your editor.

Source and Accuracy Notes

• Project page: open-and-async.com
• Source repository: github.com/open-and-async/mcp
• License: NOASSERTION — split license (code and bundled data governed separately; README confirms this)
• HN launch thread: news.ycombinator.com/item?id=48994186

What Is Open and Async MCP Server?

Open and Async MCP Server is a Model Context Protocol (MCP) server published by the team behind the Open and Async collaborative software-development playbook. It packages the book’s async-first method — originally written for remote and distributed engineering teams — as AI tools any MCP-compatible assistant can调用.

The server ships two tool groups:

Method tools (pure utility, work on any project):

Tool	What it does
draft_decision_doc	Decision + options → structured ADR/decision-doc scaffold (context, tradeoffs, decision, reversibility)
convert_meeting_to_async	Meeting purpose/agenda → async equivalent with owner and deadline
score_status_update	Scores a draft against the “work loudly / no surprises” rubric and suggests fixes
run_async_standup	Structured async-standup template a team can adopt today
triage_sync_vs_async	Recommends sync vs. async for a task, with the decision rule

Reference tools (the book’s thinking, on demand):

Tool	What it does
book_outline	Sections + chapters + one-line TL;DRs
get_chapter_summary	A chapter’s TL;DR + taglines + read link
search_principles	Keyword search over the summary corpus; short cited snippets
handle_objection	Maps skepticism (“async is slow”) to the book’s reframe
get_guidance	Role-aware (manager/ic) guidance for a topic
get_taglines	Taglines + their quote-card URLs

The server also exposes book:// resources (outline, taglines, about) and slash-command prompts: coach, async-standup, write-adr, meeting-to-issue, weekly-update.

Setup

Prerequisites

• Claude Code or any MCP-compatible AI assistant
• Node.js 18+

Install for Claude Code

claude mcp add open-async -- npx -y @open-and-async/mcp
```bash

### Install for Claude Desktop

Add to your `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "open-async": {
      "command": "npx",
      "args": ["-y", "@open-and-async/mcp"]
    }
  }
}

Any MCP client that speaks stdio works the same way — run npx @open-and-async/mcp.

Practical Evaluation Checklist

• Decision doc scaffold is structured and actionable
• Status update scoring catches surprises before posting
• Async standup template is team-ready without modification
• coach prompt composes a full deliverable in one shot
• Reference tools return cited, linked answers from the book’s corpus
• Installation completes in under 60 seconds on a fast connection

Security Notes

• Runs entirely locally via npx — no external API calls after the initial install
• No telemetry or outbound traffic beyond npm registry on install
• Split license: code is open, bundled book data has separate terms

FAQ

Q: Do I need the Open and Async book to use this? A: No. The MCP server is fully functional without the book. The reference tools draw from the book’s content, but the method tools (decision docs, standups, scoring) work standalone.

Q: Does it work with editors other than Claude? A: Any MCP-compatible AI client that speaks stdio works. The README documents Claude Code and Claude Desktop explicitly, but the protocol is standard MCP.

Q: Is the source code open? A: The code is open (MIT-compatible), but the repository has a split license — the bundled book data is governed separately. Check the LICENSE file for details.

Q: How is this different from a general-purpose MCP server? A: It is purpose-built for async-first team workflows. Rather than generic file or API access, it models collaborative practices (decision-making, status communication, meeting-to-async conversion) as structured AI tools.

Conclusion

Open and Async MCP Server brings a well-defined async methodology into the AI assistant workflow. The decision doc scaffolder alone saves significant back-and-forth on architectural choices, and the status scorer is genuinely useful for teams working across time zones. Install it with one command and it’s immediately operational — worth trying if your team does distributed or remote work.