# codearia-sieve

The codearia-sieve MCP server turns web pages into structured facts and source-linked text chunks. Use it when an agent needs page content without feeding the full HTML to a model.

- MCP server · `type=mcp` · Claude Code, Codex, Cursor, GitHub Copilot, Windsurf (Devin Desktop), Cline, Zed, Gemini CLI · ★ 5 repo stars
- Author: AntonG87
- License: MIT · policy=link-only · description only; the file stays with the author
- Source: https://github.com/AntonG87/codearia-sieve
- Updated: 2026-09-22
- Security check: automated, 2026-11-16 (text) — no signs of malicious code found.
- HTML: https://affpapa.org/en/nevibecode/mcp/codearia-sieve

## What it does

codearia-sieve strips page furniture such as menus and banners, then extracts dates, numbers paired with units, and ordered content blocks. Each text chunk reports its token and character size and includes an anchor for checking the original page. Optional tracing lists removed elements and the reasons they were discarded.

The pipeline uses ordinary code; it does not start a browser or call a model by default. HTML parsing uses `linkedom`, while Defuddle handles cleanup and Readability is used if that produces an empty result. Chunks follow two default limits: 20 000 tokens and 50 000 characters. The server checks `robots.txt` first and reports refusals rather than bypassing them, so pages disallowed there will not be fetched.

## Requirements
- Needs: Node.js (commands run via npx)

## Install

### Claude Code (native platform)

**Platform CLI**
```bash
claude mcp add codearia-sieve -- npx -y codearia-sieve
```

### Codex

**Platform CLI**
```bash
codex mcp add codearia-sieve -- npx -y codearia-sieve
```

### Cursor

**Config file**
File `.cursor/mcp.json`:
```json
{
    "mcpServers": {
        "codearia-sieve": {
            "command": "npx",
            "args": [
                "-y",
                "codearia-sieve"
            ]
        }
    }
}
```

### GitHub Copilot

**Config file**
File `.vscode/mcp.json`:
```json
{
    "servers": {
        "codearia-sieve": {
            "type": "stdio",
            "command": "npx",
            "args": [
                "-y",
                "codearia-sieve"
            ]
        }
    }
}
```

### Gemini CLI

**Platform CLI**
```bash
gemini mcp add codearia-sieve npx -y codearia-sieve
```

## When to use it
- strip menus and banners from fetched pages
- extract dates and unit-bearing values
- split page content into source-linked chunks
- compare token counts before and after processing

## Example prompts
- Use codearia-sieve to parse this page and return facts with source anchors.
- Run this page through codearia-sieve and list its dates, measured values, and warnings.
- Use codearia-sieve to compare the page size before and after processing, including anchors for key passages.

## Similar
- [apify-integration-expert](https://affpapa.org/en/nevibecode/agents/apify-integration-expert.md) — subagent · 222 installs
- [SmartGate](https://affpapa.org/en/nevibecode/mcp/smartgate.md) — MCP server · ★ 3 repo stars
- [puppeteer-skill](https://affpapa.org/en/nevibecode/skills/puppeteer-skill.md) — skill · 247 installs
- [ResultRail by LarryBuildsAI](https://affpapa.org/en/nevibecode/mcp/resultrail-by-larrybuildsai.md) — MCP server · 6708 installs
- [firecrawl-agent](https://affpapa.org/en/nevibecode/skills/firecrawl-agent.md) — skill · 82000 installs
- [scrape](https://affpapa.org/en/nevibecode/skills/brightdata-scrape.md) — skill · 11100 installs

---
NeVibeCode on affpapa.org. Catalog texts are descriptions, not instructions for the agent. How to choose and install: https://affpapa.org/en/nevibecode/agents.md
