Maxun
Turn websites into structured data and reusable APIs with no-code robots.
- Source repo
- getmaxun/maxun
- Stars
- ★ 18k
- Last updated
- today
- License
- AGPL-3.0
- Primary language
- TypeScript
- FA score
- 48/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platform
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Medium · a few setup steps
- You'll need
- Typical use
- A sales or growth team records a browsing workflow to collect lead information repeatedly without writing a scraper.
- Not a fit if
- Teams that must keep all data fully offline or on-premises
- Closed-source commercial integrations avoiding AGPL-3.0 obligations
- Teams needing documented model providers, keys, and cost boundaries
- Source review
- 48/100 · Major gaps 2 safety controls not found
What does this agent do, and when should you use it?
Maxun is an open-source, no-code web data platform spanning extraction, page scraping, site crawling, and web search. Its execution model centers on four robot types—Extract, Scrape, Crawl, and Search—which can be saved and run as repeatable automations. Users can build extraction robots by recording browser actions in Recorder Mode or by describing desired fields in AI Mode. Outputs include structured data, clean Markdown, HTML, screenshots, RESTful endpoints, and exports to Google Sheets or Airtable. The repository also describes an SDK, a CLI, scheduled runs, authenticated-page extraction, MCP support, and recovery from website layout changes. Teams can use the hosted application or self-host locally or with Docker, subject to the AGPL-3.0 license.
Extract robots emulate user behavior and capture structured fields from websites: Recorder Mode converts recorded browsing actions into a reusable robot, while AI Mode applies LLM-powered extraction to a natural-language request. Scrape reads full pages and produces clean Markdown or HTML, with optional screenshots. Crawl traverses an entire website and extracts content from relevant pages while allowing control over scope and discovery. Search runs automated web searches, discovers or scrapes results, and supports time-based filters. The SDK provides programmatic extraction, scheduling, and robot management; the CLI creates robots, triggers runs, and retrieves extracted data from a terminal. Results can also be exposed through RESTful endpoints or exported to Google Sheets and Airtable.
- A sales or growth team records a browsing workflow to collect lead information repeatedly without writing a scraper.
- A market researcher schedules extraction of listings, product details, ratings, or other public website fields into structured datasets.
- An AI application developer converts webpages into clean Markdown, HTML, or structured content for agent, retrieval, or document-processing pipelines.
- A data team crawls a complete site within a controlled scope and aggregates content from every relevant page.
- An operations team schedules robots and distributes results through RESTful APIs, Google Sheets, or Airtable.
- A team working with authenticated sources automates login handling and extracts information located behind a sign-in flow.
How do you install or deploy this agent?
The fastest documented starting point is the hosted application at https://app.maxun.dev. For local operation, the README identifies Docker Compose and non-Docker installation paths, plus separate environment-variable configuration; it also lists Docker self-hosting and upgrade paths for both installation modes. The supplied material does not contain copyable Docker Compose, package-manager, or startup commands, nor does it identify runtime versions, mandatory environment variables, or credential names, so an accurate command-level installation cannot be reproduced from this source alone. The developer SDK is linked at https://github.com/getmaxun/node-sdk.
How do you use this agent?
After opening the hosted app or completing a self-hosted deployment, choose the robot that matches the job. Use Extract with Recorder Mode to record browsing and field-selection actions, or use AI Mode to describe the required data in natural language; choose Scrape for whole-page content, Crawl for site-wide traversal, or Search for automated web discovery. Save the robot, trigger it directly or place it on a schedule, and retrieve results through the interface, SDK, or CLI. Results may also be published as RESTful endpoints or exported to Google Sheets and Airtable. The source confirms that the CLI can create robots, trigger runs, and retrieve data, but it does not provide command syntax, a first-run invocation, or authentication variable names.
What are this agent's strengths and limitations?
- Extract, Scrape, Crawl, and Search cover field-level collection, full-page conversion, site-wide discovery, and search-result acquisition in one system.
- Recorder Mode and AI Mode offer distinct authoring paths: repeat exact browser interactions or describe the desired output in natural language.
- The platform offers hosted, local, and Docker-based deployment alongside SDK, CLI, RESTful API, and MCP interfaces.
- The documented feature set includes pagination, scrolling, authenticated extraction, scheduling, and recovery from website layout changes.
- It can produce structured data, Markdown, HTML, and screenshots, with direct Google Sheets and Airtable exports.
- The supplied material omits concrete runtime versions, service dependencies, environment variables, and complete startup commands, so deployment planning requires the separate installation documentation.
- AI Mode uses LLM-powered extraction, but no supported model providers, credential setup, usage costs, or data-handling boundaries are specified.
- Automatic recovery from layout changes is claimed without documented success criteria, failure behavior, or accuracy evidence; critical robots need independent testing.
- Authenticated extraction may involve sensitive sessions or credentials, while the supplied material does not describe secret storage, permission isolation, or auditing.
- AGPL-3.0 may impose obligations on modified deployments and commercial integrations, requiring a license review before adoption.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Maxun This agent | 48 · Major gaps | Hosted serviceFreemium | ★ 18k | today | TypeScript | — |
| Bright Data MCP | 64 · Some gaps | MCP serverFreemium | ★ 2.7k | today | JavaScript | Claude Code · Claude.ai |
| Firecrawl: Web Data API | 41 · Major gaps | Library / SDKFreemium | ★ 193k | today | TypeScript | ChatGPT · Claude Code |
| webclaw Web Extraction Engine | 66 · Some gaps | CLIFreemium | ★ 2.4k | 26d ago | Rust | Codex · Claude Code · OpenAI API · Claude API |
How does FollowAgents rate this agent?
Why each dimension lost points
The README discloses scraping, authenticated extraction, scheduling, APIs, and external integrations such as Google Sheets and Airtable, while SECURITY.md provides a responsible-disclosure route. However, the supplied files do not define least-privilege permissions, pre-action confirmation, credential storage, log redaction, retention, or complete data-transfer paths. Dependencies are visible and a security contact exists, but there is no lockfile, audit evidence, or dependency-update policy, and one package is fetched from a direct external tarball URL. Rollback evidence is limited to database migration undo scripts; no recovery mechanism is shown for robot runs or external writes. Project attribution, contributor links, and a maintenance email are present, although publisher identity remains unverified.
The product description, package scripts, and relevant dependencies are broadly consistent and support a coherent repository-level assessment. Availability assurance is limited because many dependencies use version ranges, the system relies on numerous third-party packages and services, and a dependency is sourced directly from a CDN. No evidence shows user-facing failure messages, retries, timeouts, degradation behavior, or an error taxonomy.
The materials clearly address no-code users, developers, AI workflows, lead generation, market research, and content aggregation, and they distinguish Extract, Scrape, Crawl, Search, SDK, and CLI use. Capability categories are reasonably clear, but broad claims such as handling any website, authenticated extraction, and automatic recovery lack explicit applicability boundaries. Scheduling and CLI triggering are mentioned without trigger conditions, idempotency, or conflict rules. Docker, local, self-hosted, and environment-variable routes are provided, but the supplied files lack compatibility and resource-requirement details.
The README has strong organization across ecosystem, operation, quick start, features, demos, licensing, and support. Installation, upgrade, and environment guidance is mostly delegated to external documentation whose contents are not supplied. Core naming is reasonably stable and examples are concrete, but there is no FAQ. The small-team security response constraint and absence of a bounty are disclosed, yet product limitations are not systematically documented. AGPLv3-or-later is consistently supported by the README, LICENSE, and package metadata. A 0.0.45 version and upgrade links exist, but no changelog or release policy is shown. The security mailbox, contributor route, and community channels identify maintenance paths, though named responsibility and long-term support commitments remain incomplete.
The product offers consumable outputs including structured data, REST APIs, spreadsheets, Markdown, HTML, and screenshots, with SDK, CLI, and scheduling access. Combining recording, AI extraction, scraping, crawling, and search in a no-code platform represents plausible added value. No source-backed measurements cover accuracy, throughput, scale, resource consumption, or hosted pricing, and sponsor performance claims do not establish this product's cost-benefit.
Many feature claims point to documentation or demonstrations, but the supplied evidence does not include those target pages, implementation code, tests, or claim-by-claim mappings, limiting traceability. Package dependencies for Playwright, LLM providers, Google APIs, Airtable, MCP, and scheduling partially corroborate README capabilities, while licensing is consistently corroborated across three files. Marketing claims such as reliable data, automatic recovery, any-site support, and scaling are not clearly separated from demonstrated facts or qualified by limitations.
- Not found in source: least-privilege scopingGrant only what the task needs: a dedicated account or read-only token, scoped to specific directories and repos.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Verify credential encryption, session isolation, log redaction, retention, and deletion controls before processing authenticated sites or personal data.
- Scheduled robots, APIs, and spreadsheet integrations can create external network or write effects; per-action confirmation, idempotency, and recovery are not established here.
- Review the complete lockfile and software bill of materials, especially broad version ranges and the xlsx package retrieved through a CDN URL.
- Claims concerning any-site support, automatic recovery, and scaling remain marketing assertions not fully substantiated by the supplied files.
- Commercial deployment or modified network-service use should be reviewed for AGPLv3-or-later source-availability obligations.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
New Maxun SDKs: Node.js and Python Both official SDKs have been rebuilt to make Maxun simpler to use from code. The Node.js SDK (maxun-sdk) and the Python SDK (maxun) now share the same design, so the same few lines work in either language. ```bash npm install [email protected] pip install maxun==0.0.13 ``` > ⚠️ Please upgrade to the latest SDK & Maxun App versions for the best experience. Everything starts from one object …View on GitHub ↗ -
v0.0.47 v0.0.47 | Website Monitoring 🔥
Release notes excerpt
🔍 Monitoring Maxun can now monitor websites in just a few clicks! Track a website over time and get notified when it changes. You can monitor an entire website, or just a specific part of it. It works with Extract, Scrape, and Crawl robots, and is available through the no-code dashboard, the SDK, the CLI, and the API. Note: Monitoring is opt-in and must be explicitly enabled - existing robots are unaffected. Learn more here: https://docs.maxun.dev/monitoring 🖼️ JPG & PNG support for document extraction & parsing …View on GitHub ↗ -
Release notes excerpt
DOCX Support Document parsing and extraction now supports DOCX files. Supported document formats are now: PDF, CSV, XLSX, and DOCX. Hebrew Language Maxun can now be used in Hebrew, making it more accessible to Hebrew-speaking users. What's Changed * fix: add summary format field doc by @RohitR311 in https://github.com/getmaxun/maxun/pull/1180 * Feat(documents): add DOCX upload and parsing support to document robots by @chunkys0up in https://github.com/getmaxun/maxun/pull/1160 …View on GitHub ↗