Apache-2.0, Kotlin/Rust browser automation engine for AI agents — native CLI + MCP backend, X-SQL extraction, swarm crawling, and a 50+ tool programming-agent kernel, with an explicit DeepSeek Harness integration path.
DSH integration
Compatible
Author-claimed
Safety audit
Unaudited
Last verified
2026-08-29
License
Apache-2.0
01What can it help you accomplish?
Drive a real browser inside DeepSeek Harness to navigate, click, fill, snapshot, and scrape data for an agent task
Live page snapshots, X-SQL structured data, screenshots / PDF, and autonomous agent task results
AI agents and developers building browser automation, web extraction, or large-scale crawl workflows on DeepSeek Harness
Turn downloaded HTML files into structured spreadsheets and reports without spending LLM tokens
Interactive HTML reports and Excel spreadsheets produced by WebMiner ML clustering
Data / analysis teams that need large-scale, low-cost web data structuring
02How to install into DeepSeek Harness
Prerequisites
- Node.js — required for the npm install path (`npm install -g browser4-cli`)
Installation steps
- 01
$ npm install -g browser4-cli
- 02
browser4-cli install
- 03
$ curl -fsSL https://browser4.oss-cn-beijing.aliyuncs.com/scripts/install-browser4-cli.sh | bash (Linux / macOS)
- 04
$ dsh plugin --profile web add dsh-browser4 (add the Browser4 companion plugin inside DeepSeek Harness)
Verify the integration
Not specified by the author
03DSH integration and capability boundaries
Compatible integration with DeepSeek Harness via the companion `dsh-browser4` plugin — add it with `dsh plugin --profile web add dsh-browser4`
Agent browser driving
Live pages passed through browser4-cli (snapshot refs or CSS selectors)→Page snapshots, clicks / fills, screenshots, PDF, and structured DOM extraction
Drives a real Chrome / Chromium browser sessionWrites screenshots / PDFs and session state to diskZero-Token HTML→Spreadsheet extraction
Downloaded HTML files / crawled HTML directories→Structured Excel spreadsheets and interactive HTML reports via WebMiner ML clustering
WebMiner requires JDK 17+ (auto-detected)Clustering runs locally with no LLM tokensSwarm crawling & batch automation
Seed URL lists or seed files→Parallel crawl / swarm jobs with X-SQL extraction results
Launches parallel browser contexts / processesNetwork egress to crawled sitesProgramming-Agent Kernel (browser4-coding)
Agent tasks that build Browser4 artifacts or Browser4 itself→50+ `coding.*` tools: sandboxed shell/fs, scaffolding, validation, self-development
Executes shell commands in a sandbox (command whitelist)Writes / validates files on disk
04Who is it for? When not to use it?
Good for
- AI agents and developers building browser automation, web extraction, or large-scale crawl workflows on DeepSeek Harness
- Data / analysis teams that need large-scale, low-cost web data structuring
05Compatibility, maintenance and safety notes
- AI-powered commands (`extract`, `summarize`, `chat`, `agent run`, and X-SQL `llm_*` functions) require an LLM provider key such as `DEEPSEEK_API_KEY`.
- Browser4 runtime downloads / install operations and crawl / swarm jobs require network access; a proxy can be configured via `PROXY_ROTATION_URL`.
- WebMiner (zero-token spreadsheet extraction) requires JDK 17+ (auto-detected).
- Building from source requires JDK 25+ (Eclipse Temurin), Chrome/Chromium, and PowerShell 7 (Linux/macOS only).
Apache-2.0 · actively maintained (latest release v4.13.11, 2026-08-25)
06Frequently asked questions
How do I connect Browser4 to DeepSeek Harness?
Browser4 ships a companion plugin. From the README: run `dsh plugin --profile web add dsh-browser4` (npm registry) or `dsh plugin --profile web add github:platonai/dsh-browser4` (GitHub). You can also install the `browser4-cli` directly and point it at DeepSeek Harness.
Do the AI-powered commands need an API key?
Yes. Commands like `extract`, `summarize`, `chat`, `agent run`, and X-SQL `llm_*` functions require an LLM provider key — for example `DEEPSEEK_API_KEY`. Pure browser interaction and X-SQL extraction over stored HTML need no key.
Can I extract data without spending LLM tokens?
Yes. Use `htmlsnapshot query --sql @query.sql` for deterministic DOM extraction, or `htmlsnapshot readability` for article text — no tokens. For turning many HTML files into spreadsheets, `browser4-cli webminer` runs WebMiner ML clustering locally with no LLM cost (needs JDK 17+).
How do I run large-scale crawls?
Use `crawl` from a URL or seed file, or `swarm create` + `swarm query --seed-file ...` for parallel execution. The engine is CDP-native and built for 100k–200k complex page visits per machine per day via swarm/crawl scale-out.
What is the Programming-Agent Kernel?
The `browser4-coding` kernel exposes 50+ `coding.*` tools (sandboxed shell/fs, scaffolding, validation, self-development) so an agent can build Browser4 artifacts — or develop Browser4 itself — independent of the agent backend.
07Related DSH workflows
last30days-skill-cn
by jesseovo
AI agent skill that searches China's eight major platforms for the last 30 days of content and compiles a well-sourced research report.
dsh-browser
by lum1104
让 DSH 读取/点击/填表/滚动/导航当前 Chrome 标签页（文本优先，保留登录态）
anysearch-dsh
by anysearch-team
AnySearch web search provider and advanced search tools for DeepSeek Harness (DSH)
modsearch
by liustack
Web/X 搜索与抓取桥：为纯文本 coding agent 提供结构化 JSON 证据（search/fetch/citations），引擎可选 Antigravity CLI（免 key）/带 key 引擎/X（Grok Build 登录态）。
08Data and sources
https://github.com/platonai/dsh-browser4
dsh plugin --profile web add dsh-browser4
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-29. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.