Back to vendors
D

Decodo

Also known as: Smartproxy, Smartproxy.com

Visit site
Entry priceFree plan, paid plans from $19 a monthFull pricing detail

Decodo, formerly Smartproxy, supplies live public web data to developers, data teams and AI agents. Its Web Data API covers page extraction, site discovery, bulk crawling and search. Cursor or Claude Code can use the same service through MCP, and an application can call it through the API or a client library. Decodo handles JavaScript rendering and returns content in formats such as Markdown or structured data. A free monthly allowance supports initial and recurring small jobs.

Decodo supplies web data and proxy infrastructure to developers, analytics teams and AI agents. The company traded as Smartproxy until April 2025 and now sells a Web Data API that combines four operations under one account. Scrape retrieves a page, Map discovers URLs, Crawl finds and collects a set of pages, and Search returns structured search results. The API can turn a page into Markdown for an assistant or return HTML for an application that needs the original structure.

Developers can use the API directly, install a client library for TypeScript, Python or Go, or invoke the CLI from a shell. Beneath the API sits a proxy network of more than 125 million IPs, with residential addresses in more than 195 locations, sold separately as residential, mobile, ISP and datacenter proxies. Decodo counts more than 135,000 clients, among them InfoPrice, GobbleCube, InSites, ExoClick and AdSecure, and Skaler.app uses it to collect data from more than 10 million ads a month.

A team building an assistant can add Decodo's MCP server to an existing client and let the assistant retrieve the sources it needs while answering a question. Integrations for LangChain, LlamaIndex, n8n and OpenClaw give retrieval applications other ways to use the service. An engineer building a public site index can ask Crawl to discover links and collect up to 10,000 pages in a job. The service manages the queue and page retries.

Crawls run as background jobs and can notify the developer's application through a callback when the results are ready. Decodo offers more than 100 scraping templates, with targets including Shopify, Bing and Amazon, and the agent skills package helps a coding assistant select the appropriate operation and output format. Response times run under 0.2 seconds, with a 99.99% success rate on hard targets such as Amazon and sites behind Cloudflare.

The free monthly Web Data API plan includes $5 of request value, which covers up to 10,000 pages at the standard scrape rate. Paid plans begin at $19 a month and raise the allowance and, at higher tiers, the request rate, with a 14 day money back guarantee. Decodo charges for successful results. On the $19 plan, 1,000 standard scrape requests cost $0.50 against the allowance, or $0.75 with JavaScript rendering. Search is $0.75 per 1,000 on that tier, and the lowest listed rate is $0.15 per 1,000 requests. Crawl charges separately for each page it returns. Decodo holds ISO/IEC 27001:2022 certification.

Decodo fits teams that need to give an existing agent current public sources or collect a repeatable data feed for research, search visibility and pricing analysis. A research agent can run a search for source candidates, fetch the relevant articles and pass the Markdown to its own model. An engineering team can also crawl a public documentation site and load the exported pages into a retrieval index it operates. The managed service handles page collection and formatting. The application that calls it decides what work to perform with those results.

Vendor details

Canonical URL

https://decodo.com

Category

Agent infrastructure

Subcategory

Managed web data retrieval for agents, engineering teams and data pipelines

Funding status

Operated by UAB Data troops in Vilnius, Lithuania.

Company status

independent

Use cases & customers

Primary use cases

Live web retrieval for assistantsBulk public documentation crawlSearch results and public catalog data collection

Target customers

Developers building research agentsData engineers managing public web data pipelinesMarket intelligence and search teams

Deployment options

Decodo hosted Web Data APILocal SDK, CLI or MCP client connected to the hosted service

Integrations

MCP for coding assistants and other clients, client libraries for Python, TypeScript and Go, plus LangChain, LlamaIndex, n8n and OpenClaw.

In practice

A research assistant searches the public web for sources about a market and retrieves Markdown copies of the pages it selects.

An engineering team crawls a public documentation site and sends the results into its own retrieval datastore after receiving the job completion callback.

A search team collects structured results for a recurring set of queries and compares which public pages appeared across reporting periods.

Agentic Index coverage score

5.5 / 14 capabilities · 39%

Integrations & Tool Calling Partial

The MCP server supplies public web content to the caller. It can search, discover links, fetch a selected URL or return a crawl result, and the buyer can connect the same service through LangChain or LlamaIndex. These connections bring external data into the caller's task. The tools retrieve pages and search results without writing changes back to the websites they read.

Sourcedecodo.com/web-data-apiread 2026-11-20

Workflow Orchestration Partial

Crawl combines discovery and page extraction in one managed job. It collects links, applies the customer's filters, queues pages and retries pages where the initial collection fails. A job can cover up to 10,000 pages. The application controls any follow up work, such as scheduling another crawl or feeding the collected materials to a separate processing pipeline.

Sourcedecodo.com/web-data-api/crawlread 2026-11-20

Knowledge Grounding & RAG Partial

An agent can retrieve current source material by searching the web, discovering site links or retrieving the text of a page. Markdown keeps that page ready to place in the caller's context, while parsed results return fields the caller can use directly. For a maintained knowledge base, the buyer loads the collected material into a retrieval index outside Decodo.

Sourcedecodo.com/web-data-apiread 2026-11-20

Human Oversight & Guardrails Not documented

A person authorizes the connection in the MCP client before an agent starts retrieving public pages. Later tool requests use the agreed connection to return search or extraction results. The caller can set the scope of a crawl by providing filters and limits. Review and approval of the agent's individual next steps are handled in the client that invokes those tools.

Sourcedecodo.com/ai/mcp-serverread 2026-11-20

Security, Identity & Governance Partial

Decodo holds ISO/IEC 27001:2022 certification and offers a data processing agreement. Web Data API requests use an account credential for the endpoints that collect web content, and the company's access controls sit inside that certified security program. The API has no separate operator roles or permissions for a team's retrieval jobs.

Sourcetrust.decodo.comread 2026-11-20

Observability & Auditability Partial

A developer can monitor whether a background crawl is still running and collect the individual pages as the job produces them. Completion can be sent to a webhook, and the result stays available for roughly 24 hours. Job status and the returned pages explain the state of the collection. Decisions made by the agent that requested it stay in the calling application.

Sourcedecodo.com/web-data-api/crawlread 2026-11-20

Memory & State Persistence Not documented

A crawl can return its pages while it runs. The hosted job retains the output for roughly a day, so an application can collect it after execution, and a callback can signal completion. The retained pages are results of that specific job. The calling agent manages any context it carries into the next request, including summaries of the sources it already gathered.

Sourcedecodo.com/web-data-api/crawlread 2026-11-20

Deployment & Data Residency Not documented

A developer uses one account to reach the managed retrieval service from an application, a local terminal or an assistant. The client runs where the buyer places it and the page collection runs on Decodo's infrastructure. Geographic targeting can change the market from which a page is collected. It does not change where the managed service processes its job data.

Sourcedecodo.com/web-data-apiread 2026-11-20

Prebuilt Agents / Templates / Packs Full

Prepared templates turn public search pages and storefronts into records for recurring data collection jobs. Available targets include Bing, Shopify and Amazon, with templates also covering search suggestions and real estate listings. Decodo's agent skills give a coding assistant packaged instructions for picking the appropriate operations and output formats.

Sourcedecodo.com/web-data-apiread 2026-11-20

Triggers & Channel Coverage Partial

A crawl starts when a client submits a job and can notify a supplied webhook when it finishes. The callback can wake the caller's own data pipeline as soon as results are ready, and polling is also available. Searches and individual page requests run on demand. The initiating application controls schedules and signals that create new collection work.

Sourcedecodo.com/web-data-api/crawlread 2026-11-20

Model Flexibility & Routing Not documented

A compatible assistant chooses which Decodo retrieval tool to call and what to do with the pages returned. Model choice for the assistant belongs to that client's own configuration. The Web Data API options select the data operation and output format. The request surface has no model picker for the hosted components that prepare search results and extracted pages.

Sourcedecodo.com/ai/mcp-serverread 2026-11-20

APIs / SDKs / MCP Extensibility Full

The Web Data API is available directly for applications that fetch a page, map URLs, collect a site or search the web. Developers can use a client library in TypeScript, Python or Go, or run the command line tool from a shell or job. The MCP server exposes Decodo's retrieval operations to agent clients over the same account and its allocated request value.

Sourcedecodo.com/developersread 2026-11-20

Testing, Debugging & Optimization Not documented

A developer can issue a test request from a terminal and inspect the response and its format before wiring it into an application. The free monthly allowance supports checks with sample sites and queries. These calls test whether retrieval works for the intended source. The service does not score how a calling agent behaves or compare the quality of different agent versions.

Sourcedecodo.com/developersread 2026-11-20

Browser / Computer-use Partial

Public page extraction can return HTML or Markdown to an agent and can render JavaScript when the site needs it. Site mapping discovers URLs, while crawling collects content from a group of those URLs. The caller gets web content and search results from the managed retrieval tools. It has no interactive browser session for navigating a site or operating its controls.

Sourcedecodo.com/web-data-apiread 2026-11-20

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Pricing

Free plan, paid plans from $19 a month

Monthly request value and endpoint rates

Free tier

What is public

Decodo charges $19, $49, $99, $249, $499, $999 or $1,499 a month for the Web Data API, with a free allowance of $5 that covers up to 10,000 pages. The $19 plan allows 10 requests a second and covers up to 38,000 standard scraped pages, and the $1,499 plan covers up to 9.9 million. Higher plans reach 200 requests a second, and the lowest listed rate is $0.15 per 1,000 requests. Enterprise is quote only.

Billing mechanics

On the free and $19 plans, 1,000 successful standard scrape requests use $0.50 of the allowance without JavaScript rendering or $0.75 with rendering. Premium scrape requests use $1 or $1.50 respectively. Search uses $0.75 per 1,000 results requests. Crawl uses the premium rate for each page that returns.

Variable cost rationale

Each plan carries a fixed request value, and the endpoint, proxy tier and rendering choice decide how quickly a job spends it.

Additional watchouts

Scrape requests default to the premium proxy rate. A Map call can use one request for the submitted page and a second for its sitemap. Allowance examples vary with the endpoint, proxy tier and rendering choice.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free recurring $5 allowance

Lowest paid plan

Web Data API $19 monthly plan

Commercial notes

Plans renew monthly and carry a 14 day money back guarantee. Proxy products bill separately, from $2 per GB for residential traffic.

Missing data

Enterprise contract pricing is quoted.

Agentic Index verified 2026-11-20

Alternatives to Decodo

Other products that do the work Decodo does, and what each one is.

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.