Skip to content

DOS AI MCP

Remote MCP server (Streamable HTTP, Bearer API key) — DOS AI is a platform for AI assistants (chatbots) in WhatsApp and Telegram with a built-in CRM. The MCP endpoint exposes the same data a human operator sees in the cabinet: projects, conversations, messages, leads, analytics and balance — through 13 tools that carry no destructive operations by design.

Server type: Remote (Streamable HTTP)
Auth: Bearer API key (dos_sk_live_...)
Endpoint: https://dosai.pro/api/mcp
Tools: 13 (dosai_-prefixed, no destructive tools)
Pricing: starting balance granted on signup, no card required
Category: Communication
Built by: DOS AI (dosai.pro)

Why This Matters for Operators

Chatbots on WhatsApp and Telegram are only useful if someone operates them: watching conversations, catching leads, correcting prompts, and handing the dialog to a human when the rules say so. The DOS AI MCP puts that operator loop in the agent itself. A CorpusIQ-style agent can read the conversations, file leads, update a bot's prompt, send an operator message into a live dialog, and pull conversion analytics — from any MCP client.

Two design decisions make this safe to hand to an agent: there is no delete-project, no payment, and no member management — the destructive tool list is empty and a test guards it — and every tool calls the public REST route over HTTP, so a read-only key stays read-only and project scoping applies exactly once. Keys can be scoped to a single project and to read-only access, and only a SHA-256 hash of the key is stored.

The OpenAPI spec, llms.txt, and a full public corpus (llms-full.txt) are published — agent-readable documentation as a first-class deliverable.

Tools & Capabilities

Tool Purpose
list_projects / get_project Bot projects and their configuration
get_prompt / update_prompt Read and revise a bot's instructions
list_functions Functions the bot can call
list_leads / get_lead / update_lead CRM leads with status and source
list_conversations / get_messages Dialogs with filters and paging
send_operator_message Interject into a live dialog as the operator
get_analytics Dialogs, leads, conversion, response time
get_balance Account balance and token history

Installation

# Register at dosai.pro, create a project, issue an API key in the cabinet
claude mcp add --transport http dosai https://dosai.pro/api/mcp \
  --header "Authorization: Bearer dos_sk_live_..."

Rate limit is 120 requests per minute per key with Retry-After on 429. Webhooks deliver new-lead, new-message and stage-change events to your own endpoint with signature verification.

Configuration

{
  "mcpServers": {
    "dosai": {
      "url": "https://dosai.pro/api/mcp",
      "headers": { "Authorization": "Bearer dos_sk_live_..." }
    }
  }
}

Business Relevance

  • Chatbot operators can supervise WhatsApp and Telegram assistants from any MCP client
  • Sales teams get every conversation filed as a lead with source and status
  • Support leads can interject into live dialogs as the operator when the rules say so
  • Growth teams get conversion and response-time analytics per bot without a dashboard login

Integration with CorpusIQ

DOS AI is the messaging front-end that feeds the CorpusIQ CRM backbone. Leads captured in WhatsApp or Telegram conversations can be pushed into HubSpot or LeadConnector through the CorpusIQ CRM connector, so a chat inquiry becomes a pipeline contact instead of a lost dialog. The CorpusIQ calendar connector closes the loop the platform already starts — DOS AI bots book Google Calendar slots natively, and CorpusIQ reads the same calendar for meeting context. Analytics from DOS AI (get_analytics) can sit beside CorpusIQ GA4 in cross-source reporting: bot dialogs versus site traffic, one attribution view.

Limitations

  • Brand new — no track record yet; listed August 17, 2026
  • WhatsApp/Telegram only — no web or other messaging channels
  • No destructive tools means deletions still happen in the human cabinet
  • Rate limits (120 req/min per key) require pacing for heavy analytics pulls
  • Hosted platform — conversation data lives on DOS AI infrastructure; webhooks help you take it out

See Also