# Nougat OCR

> Use this tool when you need to accurately extract equations, tables, and text from scientific PDFs, and require high-fidelity OCR output in formats like raw Nougat or markdown with KaTeX compatibility. It solves problems of inaccurate text extraction and formatting issues in academic documents, providing reliable output for agent workflows. Ideal for use cases involving dense academic layouts, where precise rendering of mathematical equations and tables is crucial.

Canonical page: https://skillsregistry.net/skills/svretina-nougat-ocr  
JSON: https://api.skillsregistry.net/v1/skills/svretina-nougat-ocr

## Description

Provides high-fidelity OCR of scientific PDFs using Meta's Nougat model, purpose-built for agent workflows that require accurate extraction of equations, tables, and dense academic layouts. Supports two output formats: raw Nougat/Mathpix-style output and a renderer-friendly markdown conversion with KaTeX compatibility fixes. Configurable via a settings file so agents can share a default format policy across sessions.

## Trust

- **Trust score (0–1):** 1.00
- **Verification tier:** verified
- **Last scanned:** 2026-09-28

## Facts

- **Version:** 1.0.0
- **Skill type:** atomic
- **Execution layer:** mcp-remote
- **Runtime environment:** api
- **Category:** file-system
- **Updated:** 2026-09-28

## Source

- **Source listing:** [PulseMCP](https://www.pulsemcp.com/servers/svretina-nougat-ocr)
- **Repository:** <https://github.com/svretina/nougat-mcp>

## Use it

Resolve this record through the SkillsRegistry MCP server (no auth, read-only):

```
claude mcp add --transport http --scope user skillsregistry https://api.skillsregistry.net/mcp
```

```json
{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "tools/call",
  "params": {
    "name": "get_skill",
    "arguments": {
      "slug": "svretina-nougat-ocr"
    }
  }
}
```

REST: `GET https://api.skillsregistry.net/v1/skills/svretina-nougat-ocr` · pull for local use: `GET https://api.skillsregistry.net/v1/skills/svretina-nougat-ocr/pull`

---
SkillsRegistry indexes agent skills from public registries and GitHub. Skills we have analysed are scanned with Circle-IR and scored on six dimensions; each listing states its scan coverage. More: https://skillsregistry.net/llms.txt