doQment

Product Introduction

1. doQment is a MCP-as-a-service platform designed to convert any website into a structured information retrieval system optimized for AI coding agents. It enables developers to create ephemeral, domain-specific MCP servers through automated or manual crawling, eliminating the need for UI-based interactions.
2. The product’s core value lies in providing AI agents with instant access to clean, tool-ready website data through scoped API endpoints, while minimizing integration overhead and infrastructure costs. It prioritizes agent-centric workflows by exposing search, retrieval, and page navigation as native MCP tools.

Main Features

1. The platform automatically transforms crawled websites into structured corpora containing pages, images, forms, tables, and interactive elements, formatted for direct use by AI agents through JSON-RPC endpoints. Crawls preserve DOM hierarchy and metadata while filtering irrelevant content for machine consumption.
2. Each crawl generates a unique MCP connection URL with time-limited authentication, allowing agents to immediately query data via SSE (Server-Sent Events) or REST without SDK integration. Connections include scoped API keys and auto-expire 2-6 hours after crawl completion.
3. Developers initiate crawls through a single REST API call (POST /api/v1/domains/crawl) with configurable modes: DEFAULT (200-page limit, depth≈5) or MAX (unbounded depth with subdomain inclusion). Real-time crawl status is accessible via JSON-RPC or dashboard monitoring.

Problems Solved

1. The product eliminates manual data preprocessing and complex API integrations required for AI agents to interact with website content, reducing setup time from hours to minutes. Traditional web scraping methods often fail to provide structured, tool-native data formats suitable for autonomous agent operations.
2. It specifically targets developers building AI-powered coding assistants (e.g., Claude Code, Cursor IDE plugins) that require real-time access to documentation sites, knowledge bases, or dynamically updated web resources.
3. Typical use cases include creating temporary MCP instances for agent-based code documentation lookup, automated form interaction testing, or time-sensitive research tasks requiring structured web data extraction without persistent infrastructure.

Unique Advantages

1. Unlike conventional web scraping services, doQment provides MCP-compliant endpoints that expose website interactions as tool functions (e.g., search_page(), extract_table()), compatible with major AI agent frameworks without middleware. Competitors typically return raw HTML or unstructured JSON.
2. The platform implements automatic connection expiration and scope-limited API keys per crawl session, ensuring security for transient agent operations. Most alternatives require manual session management or provide permanent access credentials.
3. Competitive pricing at $0.001 per crawled page and $0.0001 per tool call enables cost-effective experimentation, particularly for AI agent developers needing multiple short-lived MCP instances. The prepaid credit system with auto-recharge thresholds prevents service interruptions during active agent operations.

Frequently Asked Questions (FAQ)

1. How long does a typical crawl take? The DEFAULT mode processes approximately 200 pages in 15 minutes using breadth-first crawling with depth prioritization. MAX mode duration varies exponentially based on website complexity, with status updates provided through JSON-RPC crawl/status calls.
2. When does the MCP connection expiration timer start? The auto-expiration period begins immediately after crawl completion, with sessions typically lasting 2-6 hours depending on crawl size. Users can monitor remaining time through the MCP endpoint’s /status method or dashboard interface.
3. What constitutes a "tool call" in pricing? Each API interaction through the MCP connection counts as one tool call, including page retrievals (get_page), semantic searches (search), and structured data extraction (extract_table). Bulk operations using the batch_query endpoint count as single calls.
4. Can humans access the crawled data? While optimized for AI agents, users can inspect crawled content through the dashboard’s raw JSON viewer or append ?pretty=1 to API responses for formatted output. Human interaction is not required for agent integration.
5. How are subdomains handled in different crawl modes? DEFAULT mode excludes subdomains unless explicitly specified in the domain parameter. MAX mode automatically includes all subdomains matching the root domain’s TLD, with configurable allow/block lists available through advanced API parameters.