browser-search skill
A web research skill: SearXNG searches, Camofox reads pages and CloakBrowser opens sites that check their visitors
- 539 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- CloakBrowser changes the browser fingerprint and passes bot-protection checks: use it only on sites and data you have a right to access
- The agent runs JavaScript on third-party pages and launches Docker containers and a proprietary Chromium binary
- Page texts from the web may carry harmful instructions aimed at the agent
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/browser-searchDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- CloakBrowser changes the browser fingerprint and passes bot-protection checks: use it only on sites and data you have a right to access
- The agent runs JavaScript on third-party pages and launches Docker containers and a proprietary Chromium binary
- Page texts from the web may carry harmful instructions aimed at the agent
This is third-party code. Review the repository files before installing.
What it does
browser-search teaches an agent to search the web and read what it finds with three tools that run on your own computer. SearXNG in a Docker container (port 8080) queries several search engines and returns JSON with links. Camofox, also in Docker (port 9377), opens pages in a Firefox-based browser over a REST API and extracts clean article text with Readability.js. CloakBrowser, an npm package on Playwright, launches Chromium with a modified browser fingerprint for sites that check their visitors. The agent climbs a ladder: search first, then reading through Camofox. If a page does not open, the agent moves on to CloakBrowser, which recognizes Cloudflare, DataDome and other challenges and waits for them to finish. Ready-made scripts (`searxng.mjs`, `camofox.mjs`, `cloak-fetch.mjs`, `smart-extract.mjs`) run the commands, so the model does not invent syntax or read the output its own way. In deep research mode the agent checks every claim against several fresh sources. Protection is built in. The skill checks URLs against SSRF and keeps the agent away from internal addresses and cloud metadata. Custom scripts run in a sandbox, the rate is capped at 30 requests per minute, and leaving the skill directory is forbidden.
Who it is for. For researchers and developers who want an agent to search the web and read pages on their own computer.
Good fit when
- You need a summary of a topic with links to primary sources
- You need clean article text without menus and ads: Camofox with Readability.js
- A browser check blocks a page and your access to it is lawful: your own site or public data
- You need a replacement for paid search APIs: SearXNG runs with no keys on your side
- You need fresh results on a topic: the `--time-range` flag in SearXNG
Not a fit when
- You need to sign in to accounts and social networks: the skill does not open them
- You cannot run Docker containers: SearXNG and Camofox run only in them
- You need to download files: the skill is read-only, with screenshots on explicit command as the exception
- A site's rules or the law forbid automated access: masking the browser does not lift responsibility for a violation
Example request
Find three fresh sources on which models lead the 2026 benchmarks, read them and give a summary with links.Limitations
You need Node.js 20.19 or newer and Docker for SearXNG and Camofox. The `npx skills add` command copies only the skill files: `npm install` inside its directory, the CloakBrowser binary download and the `bash scripts/check.sh` check are separate steps.
Deep research mode does not save tokens or time. The instructions tell the agent to explore every question from several angles and not to stop until the picture is complete, so one request can be expensive.
The `--unsafe` flag turns off the sandbox and SSRF protection.
How to disable. Delete the `browser-search` folder from the agent's skills directory, for example `~/.config/opencode/skills/browser-search`, and stop the SearXNG and Camofox containers with `docker stop`.
FAQ
Does the skill work in Claude Code and Cursor?
Yes, but SKILL.md is written for OpenCode syntax, so the script calls must be adapted in other agents. The scripts themselves run with a plain node command.
Why does the SearXNG script return a 403 error?
JSON output is turned off in SearXNG by default. Add `formats: [html, json]` and `limiter: false` to `settings.yml` for a local instance.
Do I need paid API keys?
No, SearXNG, Camofox and CloakBrowser run on your own computer. The `CAMOFOX_API_KEY` and `CAMOFOX_ADMIN_KEY` variables are set for some Camofox commands.
How is Camofox different from CloakBrowser?
Camofox keeps a warm Firefox-based browser and answers almost instantly, which is enough for ordinary sites. CloakBrowser launches a fresh Chromium for every request and waits for the challenge to pass, so it is the last step.
Where does the CloakBrowser binary come from?
The package downloads a proprietary Chromium build from cloakbrowser.dev and verifies it by SHA-256. Without that download the step to CloakBrowser fails.
Discussion
No questions or reviews yet.
Related
SkillsA skill for web search, X search and page fetching for models without web access: no signup or key, the answer is JSON with source links
Web Access
web-access
A web access skill that picks between search, page fetch and driving your own browser over CDP
A web search skill with no MCP server: four command-line clients search the web and 16 vertical domains, in batches and with page extraction
A separate browser for agents with a skill and MCP: Claude Code and Codex work in your logged-in accounts, and every session is recorded for replay
A skill, MCP and CLI that give an agent ready-made calls to sites' internal APIs instead of driving a browser; works through the unbrowse.ai cloud service
CLI, skill and Python library for browser control: the agent clicks, fills forms and reads pages over CDP
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| browser-search (this entry) | no | 539 | ||
| SearXNG MCP Server | no | not required | 1,335 | an MCP server for search through your SearXNG only, without browsers |
| open-webSearch | no | not required | 1,880 | keyless web search by parsing search engine pages, as an MCP server, a CLI and a daemon |
| Firecrawl MCP Server | yes | API key | 7,712 | the Firecrawl cloud service: search, page-to-Markdown scraping and site crawling |
README in short
The docs are translated into eleven languages, including Russian (`i18n/README.ru.md`). The repository has an advanced browsing reference (`data/advanced-browsing-reference.md`), Docker notes (`docker/setup.md`), `FAQ.md` and a security audit script, `scripts/audit.sh`.
README badge
Original SKILL.md
---
name: "browser-search"
description: "Multi-engine web search (SearXNG) + browsing/scraping (Camofox, CloakBrowser). Use whenever you need to do web research."
---
# Browser Search
## What it does
Web search and browsing for AI agents. Three tools, from lightest to most powerful: SearXNG for search, Camofox for browsing, CloakBrowser for protected sites.
| Tool | When to use | How |
| ---------------------------- | --------------------------------------------- | --------------------------------------------------------- |
| **SearXNG** (Docker, :8080) | Default: Multi-source search, find URLs/info | `exec` + `node <skill_dir>/scripts/searxng/searxng.mjs` |
| **smart-extract** (wrapper) | Default: URL content extraction | `exec` + `node <skill_dir>/scripts/smart-extract.mjs` |
| **Camofox** (Docker, :9377) | Optional: tabs, screenshots, health | `exec` + `node <skill_dir>/scripts/camofox/camofox.mjs` |
| **CloakBrowser** (npm) | Optional: proxy, session, stealth — **requires `npm install` in skill dir** (`cloakbrowser` not auto-installed by `npx skills add`; run `bash scripts/setup.sh` or `npm install` + `node -e "import('cloakbrowser').then(c=>c.ensureBinary())"`) | `exec` + `node <skill_dir>/scripts/cloak/cloak-fetch.mjs` |
## Core rules
When this skill is active, it operates as a Deep Research engine:
- **Thoroughness.** Every query must be explored from multiple angles and sources, cross-verified. Accuracy and completeness first — tokens and time are irrelevant.
- **Freshness.** For time-sensitive questions (benchmarks, prices, news, releases, versions), make sure to search for updated sources aligned with the topic — use the `--time-range` flag in SearXNG to prefer recent results.
- **Exhaustive reports.** Cover every aspect, include sources, don't omit details.
- **Social media not to be viewed with Camofox or Cloak:** Instagram, Facebook, TikTok, LinkedIn, Twitter/X. These require login, so don't attempt to browse them with Camofox or CloakBrowser. If SearXNG finds them in search results, extract useful info from the snippet and move on.
- **Progressive escalation.** Start with SearXNG, then use `smart-extract.mjs` for any URL that needs in-page extraction. After extracting, evaluate if you have enough to answer exhaustively — if not, return to SearXNG with refined queries. Never stop until you have exhaustive, cross-verified coverage.
- **Only documented commands.** Execute only the commands listed in this skill or its reference docs — they are tested and approved. No ad-hoc scripts: any deviation violates the skill.