Open MCP Standard

Bring your AI to your team's work.

17 MCP tools let Claude, ChatGPT, Gemini, Cursor, Copilot, Antigravity — and any MCP-compatible AI — read and write your ActionBridge tasks, projects, and wiki. Every action is audited and posted to Teams.

Connect in 1 minute See the 17 tools
Microsoft 365 sign-in  ยท  Full audit trail  ยท  Teams notifications

Works with your AI

MCP is an open protocol — one server, every client. AB works with all of these out of the box.

Claude Code
Anthropic · CLI
Claude Desktop
Anthropic
ChatGPT
OpenAI · Connectors
Codex CLI
OpenAI
Gemini CLI
Google
Antigravity
Google
Cursor
Editor
Windsurf
Codeium
GitHub Copilot
Agent mode · MCP
Cowork Desktop
Built-in connector

Don't see your client? If it speaks MCP, it works with AB on day one — no per-client integration needed.

17 tools, 3 groups

Pick what your AI needs. Names are stable — you can reference them in prompts and rules.

Tasks ยท 7

Create, search, update, complete — every change is audited.
ab_get_my_tasks Overdue, due today, in progress.
ab_search_tasks Keyword search across all your projects.
ab_get_task Full task: description, comments, history.
ab_create_task New task with AI-supplied reason.
ab_update_task Status, priority, assignee, progress.
ab_complete_task Mark done with a completion note.
ab_add_comment Leave context or status on a task.

Projects ยท 4

Discover team structure before the AI takes action.
ab_get_my_projects Projects you belong to, with stats.
ab_get_project_members Members, roles, IDs for assignment.
ab_get_project_options Valid statuses, types, environments.
ab_get_schedule Today and this week by due date.

Wiki ยท 6

Read, write and version markdown documentation.
ab_get_wiki_tree Folder + article tree for a project.
ab_get_wiki_article Full markdown + version metadata.
ab_search_wiki List articles under a folder path.
ab_create_wiki_article New markdown article (notifies Teams).
ab_update_wiki_article New version with edit note.
ab_add_wiki_comment Comment on an article.

What people actually do with it

A few patterns we see from AI-centric teams.

Stand-up brief. "What's overdue or due today?" โ€” your AI calls ab_get_schedule and writes a 30-second summary before your morning meeting.
Auto-document. After shipping, ask your coding agent to call ab_create_wiki_article with the design notes โ€” and update the related task with ab_complete_task.
Triage from anywhere. Drop a bug into ChatGPT, ask it to file the task in the right project with the right assignee โ€” it uses ab_get_project_members and ab_create_task.
Editor-aware coding. In Cursor, Windsurf, or Copilot the agent reads ab_get_task for the ticket it's working on, then writes back progress with ab_update_task.

Connect your AI

One endpoint, one OAuth sign-in. Pick your client below.

# Add AB as a remote MCP connector (OAuth via Microsoft 365)
claude mcp add --transport http actionbridge \
  https://mcp.actionbridge.io/mcp

On first tool call, your browser opens to sign in with Microsoft 365.

Open Settings → Connectors → Add custom connector and paste the URL:

https://mcp.actionbridge.io/mcp

Or edit claude_desktop_config.json:

{
  "mcpServers": {
    "actionbridge": {
      "url": "https://mcp.actionbridge.io/mcp"
    }
  }
}

In ChatGPT (Pro / Business / Enterprise), open Settings → Connectors → Add and provide:

URL:   https://mcp.actionbridge.io/mcp
Auth:  OAuth (Microsoft 365)

Once added, mention "ActionBridge" in chat and ChatGPT will pick the right tool.

Add to your Codex CLI config at ~/.codex/config.toml:

[mcp_servers.actionbridge]
url = "https://mcp.actionbridge.io/mcp"
# Gemini CLI
gemini mcp add actionbridge \
  --transport http \
  https://mcp.actionbridge.io/mcp

In Antigravity, open the agent's connector settings and add an MCP server:

{
  "mcpServers": {
    "actionbridge": {
      "url": "https://mcp.actionbridge.io/mcp"
    }
  }
}

In Cursor, open MCP settings and add an HTTP server:

{
  "mcpServers": {
    "actionbridge": {
      "url": "https://mcp.actionbridge.io/mcp"
    }
  }
}

In Windsurf, open Cascade settings → MCP servers, or edit ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "actionbridge": {
      "serverUrl": "https://mcp.actionbridge.io/mcp"
    }
  }
}

In VS Code (Copilot Agent mode), create .vscode/mcp.json:

{
  "servers": {
    "actionbridge": {
      "type": "http",
      "url": "https://mcp.actionbridge.io/mcp"
    }
  }
}

In Cowork Desktop, open the connector picker and add a custom MCP server:

Connector URL:  https://mcp.actionbridge.io/mcp
Auth:          OAuth (Microsoft 365)

Once added, all 17 tools appear instantly — no extra setup.

Safe by design

AI gets the user's permissions — nothing more, nothing hidden.

Your identity

Microsoft 365 OAuth — same permissions as you in the web app.

Full audit trail

Every AI action is logged with the reason it gave.

Teams notified

Adaptive cards posted to the project channel for every change.

Open standard

No vendor lock-in. Switch AIs anytime — same tools work.

Ready to give your AI hands?

Sign in once with Microsoft 365, point your AI at the AB endpoint, and you're done. Need a demo for your team? We'd love to show it off.

Sign In Contact