A Model Context Protocol server that enables interaction with WhatsApp through message retrieval, sending, and media handling capabilities
The server provides 12 tools with reasonably descriptive names and descriptions. However, there are significant gaps in schema completeness, parameter documentation, and error handling. While tools follow a clear verb_noun naming convention (search_contacts, list_messages, send_message, etc.), many parameter descriptions lack detail about constraints, formats, and dependencies. Output schemas are not explicitly documented in the source code. The tool definitions are visible and explicit (not inferred), but schema richness and parameter validation guidance are incomplete. Most tools fall in the 45-55 range; writing tools (send_message, send_file, send_audio_message) have explicit state-modification indicators but lack confirmation/dry-run patterns.
Download media from a WhatsApp message and get the local file path.
Get WhatsApp chat metadata by JID.
Get all WhatsApp chats involving the contact.
Get WhatsApp chat metadata by sender phone number.
Get most recent WhatsApp message involving the contact.
Get context around a specific WhatsApp message.
Get WhatsApp chats matching specified criteria.
Output schemas are not documented. LLMs cannot anticipate return structure or plan downstream tool calls. For example, list_messages returns results but no schema is provided showing fields like message_id, sender, timestamp, content, media_type.
Parameter descriptions lack format constraints and valid values. For example, 'after' and 'before' in list_messages say they accept 'ISO-8601 formatted string' but don't explain timezone handling or examples. 'sort_by' in list_chats accepts 'last_active' or 'name', this should be an enum in the schema.
Destructive/write tools (send_message, send_file, send_audio_message) lack confirmation or dry-run support. Agents can accidentally send messages without a chance to review. No tool annotations (destructiveHint) are visible in schema.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 60 | 2026-07-28+ | v2 |
Get WhatsApp messages matching specified criteria with optional context.
Search WhatsApp contacts by name or phone number.
Send any audio file as a WhatsApp audio message to the specified recipient. For group messages use the JID. If it errors due to ffmpeg not being installed, use send_file instead.
Send a file such as a picture, raw audio, video or document via WhatsApp to the specified recipient. For group messages use the JID.
Send a WhatsApp message to a person or group. For group chats use the JID.
Error handling is not documented. Tools do not explain recovery paths (e.g., 'If recipient is invalid, try search_contacts first'). No indication of whether errors are retryable, user-fixable, or fatal.
Parameter naming inconsistencies. Some tools accept 'jid' (get_chat, get_contact_chats), others accept 'chat_jid' (list_messages), and others accept 'sender_phone_number'. When multiple tools operate on the same resource (chats), the parameter names should be consistent to avoid LLM confusion.
Pagination parameters are inconsistently applied. list_messages, list_chats, and get_contact_chats all support 'limit' and 'page', but no total count or next_cursor is documented in output schemas. This makes it impossible for LLMs to know when pagination is exhausted.
Recipient/phone number format is underspecified. 'send_message' accepts 'phone number with country code but no + or other symbols, or a JID', but what if the user says 'Send to Jack'? There is no fuzzy lookup or name resolution. Agents must call search_contacts first, adding friction.
send_file and send_audio_message descriptions mention ffmpeg dependency and fallback behavior but do not document expected file formats, size limits, or MIME types. This leaves agents guessing what media is valid.