WhatsApp Business automation and AI assistant platform exposing 238 tools via Model Context Protocol (MCP) over JSON-RPC 2.0
This server has severe structural issues. The source code provided is predominantly documentation website code (Docusaurus React pages, curl examples, and Node.js client samples) rather than an actual MCP server implementation. No explicit MCP server registration, initialization, or protocol handlers are visible. Tools appear to be inferred from documentation examples and marketing copy rather than from a live server implementation. Descriptions are present but generic (10-50 chars, well below the 194-char baseline). Parameter schemas are listed but lack proper JSON Schema type definitions and validation constraints. No output schemas are documented. Error handling guidance is absent. The critical issue: the repository appears to be developer documentation for a WhatsApp API service, not an MCP server codebase. Without an actual server implementation visible (no src/server.ts, no tool registration handlers, no protocol compliance layer), scoring must be conservative.
Create a WhatsApp Auto workflow
Test a workflow with a dry run
Get an AI-generated summary of a WhatsApp conversation
List recent WhatsApp conversations
Get sentiment analysis for a WhatsApp conversation
Get the current status of the shop
Get AI-generated smart reply suggestions for a conversation
No MCP server implementation visible in repository. Source code is documentation website (Docusaurus) and client examples, not server code.
Tools with empty input schemas (no properties documented): get_whatsapp_outbound_queue_status, get_whatsapp_status, list_whatsapp_workflows, get_shop_status. Schema score must be 0.
All tool descriptions are under 50 characters, well below the 194-char baseline. LLMs cannot determine when/why to select tools. Examples: 'Send a WhatsApp message' (23 chars), 'List recent WhatsApp conversations' (35 chars), 'Check the status of outbound message queue' (42 chars).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 48 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 27 | - | v1 |
Check the status of outbound message queue
Get the status of WhatsApp sessions
List all WhatsApp workflows
Send a WhatsApp message
Parameter schemas lack proper JSON Schema type definitions and validation. For example, 'session_id' is marked type:integer but no min/max bounds specified. 'trigger_type' in create_whatsapp_workflow has no enum constraint defining valid trigger types.
No output schemas documented. LLMs cannot plan downstream tool calls or extract required chaining IDs. For instance, create_whatsapp_workflow should document it returns a workflow_id for use with dry_run_workflow.
No error handling guidance. Tools like send_whatsapp_message offer no recovery hints if sending fails (e.g., 'Invalid phone format. Ensure number includes country code.'). Pattern: recovery-guide missing.
Destructive operations (send_whatsapp_message, create_whatsapp_workflow) have no confirmation step or dry-run safety mechanism. Agents cannot preview consequences before execution.
Parameter descriptions are minimal (10-20 chars) and omit format/range constraints. Example: 'phone' parameter description is just 'Recipient phone number', no mention of required format (e.g., E.164 international format with country code). 'limit' lacks bounds (should specify 1-100 or similar).
No evidence of pagination support for list tools (get_recent_conversations, list_whatsapp_workflows). No offset/limit parameters or next_cursor documented. Large result sets will exceed context windows without pagination.
Tools designed for system IDs (session_id, workflow_id) require agents to perform lookups first. No tools accept human-friendly identifiers (e.g., business name, user email), missing natural-identifiers pattern.