Skip to main content
Track new features, improvements, and updates to the Nimble SDK and Web Tools. We ship updates regularly to enhance performance, add capabilities, and improve your developer experience.
Have feedback or questions about a release? Reach out to our team
​
April, 2026

​
Release Notes for April, 2026

​
New Skills

​
Business Skills for Agent Skills

Nine new verticalized skills for AI coding assistants, organized into four categories:Install via the Nimble plugin for Claude Code or Cursor.

​
New Features

​
Custom Agent Creation via API, SDK, and CLI

Create custom extraction agents programmatically. Describe what data you need, provide a sample URL, and Nimble generates a production-ready agent. Iterate with natural language prompts to refine the output. Available via REST API (POST /v1/agents/generations), Python/Node SDKs, and CLI.View Documentation

​
New Features

New links format option for both the Extract API and Agents API. Pass "links" in the formats array to extract all URLs found on the page. Links are returned as an array of URL strings under data.links.
"formats": ["html", "links"]

"data": {
  "links": [
    "https://www.example.com/about",
    "https://www.example.com/contact",
    "https://external-site.com/resource"
  ]
}
Extract Documentation | Agents Documentation
​
March, 2026

​
Release Notes for March, 2026

​
New Features

​
Agents API: Batch Endpoint

Submit up to 1,000 agent requests in a single request with POST /v1/agents/batch. Each input runs as an independent async task. Use shared_inputs to set the agent and common delivery settings, and override per item in inputs as needed.
{
  "inputs": [
    { "params": { "keyword": "iphone 15" } },
    { "params": { "keyword": "iphone 16" } }
  ],
  "shared_inputs": { "agent": "amazon_serp" }
}
View Documentation

​
New Features Breaking Change

​
Agents API: Formats Parameter

New formats parameter for POST /v1/agents/run and POST /v1/agents/async. Pass "html", "markdown", or "headers" to include additional data alongside data.parsing.
"formats": ["html", "markdown"]
Breaking change: data.html is no longer returned by default. To continue receiving HTML in the response, add "html" to the formats array.Before:
{ "agent": "amazon_pdp", "params": { "asin": "B08N5WRWNW" } }
// → data.html was included automatically
After:
{
  "agent": "amazon_pdp",
  "params": { "asin": "B08N5WRWNW" },
  "formats": ["html"]
}
// → data.html included only when explicitly requested
View Documentation

​
New Features Breaking Change

​
Extract API: New Headers Format

New headers format option. Pass "headers" in the formats array to include the HTTP response headers returned by the server. Headers are returned as a key-value object under data.headers.
"formats": ["html", "headers"]
Breaking change: Headers are no longer returned by default. Previously, data.headers was included in every response automatically. To continue receiving headers, add "headers" to your formats array.Before:
{ "url": "https://example.com" }
// → data.headers was included automatically
After:
{ "url": "https://example.com", "formats": ["html", "headers"] }
// → data.headers included only when explicitly requested
View Documentation

​
New Features

​
Batch Extract API

Submit up to 1,000 URLs in a single request with POST /v1/extract/batch. Each URL runs as an independent async task. Use shared_inputs to set common settings (render, country, cloud delivery) across the batch, and override per item in inputs as needed.View Documentation

​
Integrations

​
Integration Connectors

New connector guides for plugging Nimble web tools into popular AI frameworks and platforms:
  • Smithery — Connect via the Smithery MCP registry with AI SDK and TypeScript examples.
  • OpenAI — Use Nimble with OpenAI function calling and the Agents SDK.
  • Anthropic — Integrate with Claude tool-use API and Tool Runner.
  • Google ADK — Use Nimble with Google ADK agents via MCP.
  • LangChain — LangChain tools and retrievers for Nimble web data.

​
Documentation

​
Nimble Docs MCP

Ask questions about Nimble directly from Claude, Cursor, VS Code, or any MCP-compatible tool. The Nimble Docs MCP server exposes the full documentation knowledge base including API references, guides, and code examples, all via the search_nimble_docs tool. No API key required.Server URL: https://docs.nimbleway.com/mcpView Documentation
​
February, 2026

​
Release Notes for February, 2026

​
New Features

​
GET Agents API

New endpoints for discovering and exploring available web search agents. List all available agent templates with GET /agents or retrieve detailed information about a specific agent including its input/output schema with GET /agents/{agent_name}.
Perfect for building dynamic agent selections or understanding agent capabilities programmatically. View Documentation

​
New Features

​
Screenshot Format

New screenshot format option for the Extract API. Capture full-page screenshots as base64-encoded PNG images for visual verification, monitoring, and archival. Screenshots automatically enable JavaScript rendering (VX8/VX10 driver). View Documentation

​
Documentation

​
Documentation Updates

  • Callbacks & Delivery Guide: Comprehensive guide covering all async result delivery options - polling, webhooks, and cloud delivery to S3/GCS. Includes setup instructions for bucket permissions. View Guide
  • Integration Documentation: New guide covering all available integration and AI connectors to Nimble SDK capabilities. View Integrations

​
New Features

​
Agent Skills API

Advanced web search skills powered by Nimble Search API. Built on the open-source Agent Skills standard for cross-platform agent compatibility, connected to you Claude, Cursor and more.Features 8 specialized focus modes (general, coding, news, academic, shopping, social, geo, location) with AI-powered answer generation and smart result filtering. Try Now
​
January, 2026

​
Release Notes for January, 2026

​
Documentation

​
Comprehensive Guides

  • Quickstart Pages: New quickstart guides for all Web Tools with installation steps, basic usage examples, common patterns, and next steps. Get started in under 5 minutes.
  • Usage Documentation: Detailed usage pages for each Web Tool covering all parameters, response formats, best practices, common use cases, and limitations.
  • API Reference: Complete API reference with request/response schemas, error codes, rate limits, and authentication methods for all endpoints. Try Now
  • Account Management: New guide covering API key generation, team member management, permissions, and security best practices. View Guide
  • Integration Documentation: New guide covering all available integration and AI connectors to Nimble SDK capabilities. View Integrations

​
New Features

​
Extract API

The /extract retrieves clean HTML, text, and structured data from any webpage. Supports JavaScript rendering, stealth mode for anti-bot protection, custom parsing with CSS selectors, browser actions, screenshots, and network request capture. Try Now

​
New Features

​
Web Search Agents API

The /agent are ready-to-use web search agent extractors for popular websites like Amazon, Google, LinkedIn, and hundreds more. No coding or CSS selectors required - just provide the template name, and template inputs such as search term or product ID and get structured data instantly. Agents are maintained 24/7 by Nimble and automatically updated when sites change. Try Now

​
New Features

​
Search API

The /search performs web searches and retrieves parsed content from top results. Supports three search depth modes (lite, fast, deep) for flexible cost/quality tradeoffs, and AI-powered answer summaries from search results. Try Now

​
New Features

​
Map API

The /map perform fast URL discovery and site structure mapping. Ideal for discovering all pages on a website, extracting sitemap URLs, and understanding site architecture without deep crawling overhead. Supports subdomain filtering and exact domain matching. Try Now

​
New Features

​
Crawl API

The /crawl perform deep website crawling for comprehensive data extraction across multiple pages. Features smart navigation, data normalization, and multi-page scraping with automatic traversing handling. Try Now