standup-update-mcp

An MCP server, distributed via npm, that turns AI-chat-history exports — GitHub Copilot Chat, Google Antigravity, Cursor, ChatGPT, Claude — into standup updates, client-meeting summaries, or written status reports.

No Python required — pure Node.

Tools exposed

• prepare_digest(path, since?) — condenses a raw export (JSON, JSONL, or markdown) or a whole folder of them into a compact digest, collapsing long code diffs so the actual conversation signal isn't buried.
• get_format_template(format_name) — returns the style guide + worked example for yesterday_today_blockers, narrative (client-facing), grouped_bullets, or all.
• get_source_guide(tool_name) — export instructions for copilot_vscode, copilot_jetbrains, antigravity, cursor, chatgpt_claude_web, claude_code, or generic (copy-paste fallback).

The server does the deterministic condensing; whichever LLM is on the other end of your MCP client does the actual "read this and write my standup" reasoning.

No install needed — just point your client at npx

Most MCP clients support running a package directly via npx -y <package>, so you don't need to npm install anything globally yourself — the client does it on first launch.

VS Code + GitHub Copilot

Create .vscode/mcp.json in your project (or Command Palette → "MCP: Open User Configuration" for a machine-wide config):

{
  "servers": {
    "standup-update": {
      "type": "stdio",
      "command": "npx",
      "args": ["-y", "@navuluribalaji/standup-call-tool"]
    }
  }
}

Open the file, click the "Start" button that appears above the server entry, accept the trust prompt, then switch Copilot Chat to Agent mode and confirm the three tools show up under the tools icon.

Claude Desktop / Claude Code

{
  "mcpServers": {
    "standup-update": {
      "command": "npx",
      "args": ["-y", "@navuluribalaji/standup-call-tool"]
    }
  }
}

Cursor / Windsurf

Same mcpServers shape as above, in .cursor/mcp.json or that client's equivalent MCP settings file.

Example usage once connected

"Give me a standup update from yesterday — my Copilot chat export is at ~/Desktop/copilot-export.json"

The model calls prepare_digest, reads the digest, optionally calls get_format_template for the format you want, and writes the update directly in chat.

Local development

npm install
node src/server.js   # runs the stdio server directly

Publishing (for maintainers)

npm login
npm publish

Once published, anyone can use it with zero install via npx -y standup-update-mcp in their MCP client config — no npm install step for end users at all.

Notes

• Antigravity has no built-in export today (chats are stored as opaque .pb files) — get_source_guide("antigravity") will tell the model to ask you to copy-paste the relevant turns instead, unless you already use a community export tool like aghistory.
• Nothing here calls out to any external API — it's pure local file processing plus static reference text. No data leaves your machine via this server.