Mailtrap

Exposes Mailtrap's email sandbox API through the Model Context Protocol, enabling LLM agents and tools to programmatically query, filter, and retrieve test emails from isolated inbox environments. Implements MCP resource and tool abstractions that map directly to Mailtrap REST endpoints, allowing stateless access to email metadata, headers, and body content without managing HTTP clients directly.

Handles secure storage and injection of Mailtrap API credentials into MCP tool calls through environment variable or configuration-based authentication. Implements credential validation at initialization time to fail fast if API tokens are invalid, and transparently attaches authentication headers to all downstream Mailtrap API requests without exposing credentials in logs or tool outputs.

Discovers and lists all available sandbox inboxes associated with a Mailtrap account, returning inbox IDs, names, and configuration metadata. Implements pagination and filtering to handle accounts with many inboxes, and caches inbox list to reduce API calls for repeated queries. Enables agents to dynamically select target inboxes without hardcoding IDs.

Provides structured query tools to search and filter emails within a sandbox inbox using criteria like recipient address, subject line, timestamp range, and read/unread status. Implements query parameter validation and pagination to handle inboxes with thousands of emails efficiently. Returns email summaries with metadata (ID, sender, recipient, subject, timestamp) enabling agents to identify target emails before fetching full content.

Fetches the complete email message (headers, body, attachments) for a specific email ID, returning raw MIME content or parsed JSON representation. Handles both text/plain and text/html email bodies, and provides attachment metadata (filename, size, MIME type) without downloading binary attachment data. Implements lazy loading to avoid fetching full email bodies until explicitly requested.

Extracts and returns metadata for all attachments in an email (filename, size in bytes, MIME content-type) without downloading binary attachment data. Enables agents to verify that emails include expected attachments and validate attachment properties (size, type) without consuming bandwidth or storage for large files.

Updates email read/unread status in the sandbox inbox, enabling agents to track which emails have been processed or reviewed. Implements atomic state updates that persist in Mailtrap's database, allowing subsequent queries to filter by read status. Supports bulk operations to mark multiple emails as read in a single API call.

Deletes individual emails or bulk-clears entire sandbox inboxes to reset test state between test runs. Implements safe deletion with optional confirmation to prevent accidental data loss. Supports selective deletion (by email ID) or full inbox purge, enabling agents to maintain clean test environments without manual Mailtrap UI interaction.

Implements MCP resource and tool schema definitions that expose Mailtrap capabilities as discoverable, self-documenting interfaces. Provides JSON schema for all tool parameters, return types, and error conditions, enabling MCP clients (Claude, LLM frameworks) to understand available operations without external documentation. Implements MCP resource types for inboxes and emails as queryable entities.