What is MCP?
The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external systems and call their APIs. Once you connect your Botdog account to an MCP client, the assistant can explore your campaigns, read your inbox, build audiences, control live outreach, and wire up webhooks — all from the chat window, in plain language.What Botdog exposes
The server splits into read tools (mcp.read) and write tools (mcp.write). Reads are team-scoped; writes touch live outreach, so the assistant confirms with you first (see Scopes).
Read:
- List and fetch campaigns, and per-campaign statistics
- Pull overall, daily, and per-user analytics
- List leads in a campaign or a list, and look up a single lead
- See replies (inbox view) — including unread-only
- List and fetch lead/audience lists
- List message snippets/templates
- List connected LinkedIn sender accounts, and read an account’s daily limits + remaining quota
- See which activity types are paused on a campaign
- List and fetch webhooks
- List and fetch team members, and your team’s profile
mcp.write):
- Create and update campaigns (name + dates)
- Add leads to a campaign or to a list
- Create lists; delete leads from a list
- Update a lead’s notes and custom fields
- Pause and resume campaigns, or just specific campaign activities
- Stop a single lead’s outreach
- Create and update message snippets
- Update an account’s daily sending limits (clamped to safe maximums)
- Create, update, and delete webhooks
Supported clients
The connection works with any MCP client that supports the remote Streamable HTTP transport with OAuth 2.1 auth:- Claude Desktop (macOS, Windows, Linux)
- ChatGPT (developer mode)
- Cursor
- Claude Code (CLI + VS Code extension)