- Home
- Skills
- Research & Analysis
- Alka Search - Free Web Search & Page Reading
Works with the AI Agents you already use
Alka Search - Free Web Search & Page Reading
Search the web, read pages, and retrieve llms.txt documentation with cited sources. Requires the free Alka Search MCP server, installed and connected separately.
Free
Alka Search - Free Web Search & Page Reading
Example session with this skill installed
Find the official TypeScript documentation. Read its TypeScript Tooling in 5 Minutes guide and summarize the installation and first compilation steps with source links.
- Read your context and instructions
- Compiled the alka search
The official TypeScript documentation includes a TypeScript Tooling in 5 Minutes guide. I read that guide and found these steps:
| Step | What the guide shows |
|---|---|
| Install | For npm users, install the compiler with npm install -g typescript. |
| Create a source file | Save a small greeting function in greeter.ts. |
| Compile | Run tsc greeter.ts to generate greeter.js. |
| Add a type | Annotate the function parameter as person: string. Passing a number array then produces a type error. |
The guide explains that TypeScript checks the structure of the code and its type annotations. Its example also notes that JavaScript can still be emitted when type errors are reported, so inspect compiler diagnostics before using the output.
Source: TypeScript Tooling in 5 Minutes.
Retrieval note: DuckDuckGo supplied search results in this run; Mojeek and Brave reported blocks. The official guide was fetched successfully. These commands summarize the guide; no packages were installed and no tutorial code was executed as part of this research.
Connects securely to your tools. The creator never sees your data.
About this skill
Alka Search helps an AI agent find public sources, read the relevant pages, and answer with supporting links. Use it to check documentation, summarize an article, or research a factual question.
What it does
- Web search: Query DuckDuckGo, Mojeek, and Brave in parallel, then merge and deduplicate results.
- Page reading: Extract HTML articles as Markdown, or read plain-text pages, with adjustable output length.
- Documentation retrieval: Try a site's llms-full.txt and llms.txt files for documentation in text form.
The companion skill explains when to search or fetch directly, how to cite evidence, how to handle blocked requests, and how to treat fetched material as untrusted data.
Requirements
This is a companion skill. Install and connect the Alka Search MCP server separately using the setup guide. You need Node.js 20.18.1 or newer with npm/npx, internet access, and a local agent client that supports MCP stdio servers. Configure your MCP client to run npx --yes @alkasima/alka-search-mcp@1.0.2 using the platform-specific instructions linked below. Git and a local build are optional. Loading this skill through Agensi alone does not install the server.
The server and companion skill are free and MIT licensed. The default providers require no API key. Optional Tavily and SearXNG fallbacks require your own configuration; Tavily may carry separate costs.
Limits
Search providers may rate-limit or block automated requests. The server does not execute page JavaScript, sign into websites, bypass access controls, or parse PDFs. Search results are cached for ten minutes. Retrieved output is capped, with truncation reported. Availability depends on the upstream websites and providers.
Source and setup
The release passes 70 automated tests. The public npm install was checked with the official MCP SDK client on Windows for server initialization, all three tools, and rejection of a private URL. Individual agent applications have different setup requirements; this is not a claim that every agent application has been tested. First-run npx installation requires internet access.
Requires a separately installed local Alka Search MCP server and Node.js 20.18.1+. Search providers can block or rate-limit automated requests. No JavaScript execution, website login, access-control bypass, or PDF parsing. Search responses are cached for ten minutes; page/text downloads are bounded and output can be truncated. Individual agent applications have not all been tested. Cloud-only agents without a local MCP runtime cannot run this stdio server directly.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 29 days ago
- Free with an Agensi account
Requires
Permissions
Routine research calls the connected MCP tools. Search queries go to DuckDuckGo, Mojeek, and Brave; page requests go to the supplied websites. Read access covers the skill's setup reference. Terminal use, dependency installation, and client configuration changes are only for user-requested setup. Optional Tavily/SearXNG fallbacks use explicitly configured TAVILY_API_KEY/SEARX_URL; no account or API key is required by default. Private-address protection stays enabled for public-web research.