Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

Appearance settings
@QuartzUnit

QuartzUnit

LLM-native tools for data collection, extraction, and agent safety. 10 Python packages on PyPI.

QuartzUnit

LLM-native tools for data collection, extraction, and agent safety.

We build lightweight, composable Python libraries designed for AI agents and local LLMs. Every tool ships with a CLI, Python API, and MCP server — no cloud dependencies required.

Ecosystem

Collect          Extract          Search           Monitor          Guard
─────────        ─────────        ─────────        ─────────        ─────────
feedkit    ───→  markgrab   ───→  embgrep    ───→  diffgrab         agent-action-policy
(RSS/Atom)       (HTML/PDF/       (semantic)       (web change      agent-loop-guard
                  YouTube)                          tracking)       llm-degen-guard
                 docpick
                 (OCR→JSON)       browsegrab ───→  snapgrab
                                  (browser         (screenshot)
                                   agent)

Libraries

Package Description PyPI Tests
markgrab URL → LLM-ready markdown (HTML, YouTube, PDF, DOCX) PyPI 114
docpick Schema-driven document OCR → structured JSON PyPI 217
feedkit RSS/Atom collection with 444 curated feeds PyPI 34
browsegrab Token-efficient browser agent for local LLMs PyPI 200
snapgrab URL → screenshot + metadata, Claude Vision optimized PyPI 29
diffgrab Web page change tracking with structured diffs PyPI 89
embgrep Local semantic search — embedding-powered grep PyPI 74
llm-degen-guard LLM output degeneration detector (4-signal scoring) PyPI 55
agent-loop-guard Agent infinite loop detection (sliding window) PyPI 78
agent-action-policy Declarative action policies for AI agents PyPI 69

Total: 959 tests across 10 packages

Showcase

End-to-end examples showing how QuartzUnit packages chain together:

Project Pipeline What it does
newswatch feedkit → markgrab → embgrep → diffgrab RSS news monitoring with semantic search and change tracking
watchdeck diffgrab → markgrab → snapgrab → guard trio Web page monitoring with visual diffs and safety guards

Design Principles

  • Local-first — No API keys, no cloud. Runs on CPU or your own GPU.
  • Composable — Each tool does one thing well. Chain them via Python or MCP.
  • MCP native — Every package includes an MCP server for LLM agent integration.
  • Zero/minimal deps — Guard libraries have zero dependencies. Extraction tools use only essential libs.

Popular repositories Loading

  1. docpick docpick Public

    Forked from ArkNill/docpick

    Lightweight OCR + Local LLM → Schema-based Structured JSON Extraction

    Python 1

  2. .github .github Public

    QuartzUnit organization profile

  3. agent-action-policy agent-action-policy Public

    Forked from ArkNill/agent-action-policy

    Declarative action policies for AI agents — composable templates for safe coding, browsing, database

    Python

  4. agent-loop-guard agent-loop-guard Public

    Forked from ArkNill/agent-loop-guard

    Framework-agnostic agent loop detection — sliding window similarity scoring, zero dependencies

    Python

  5. browsegrab browsegrab Public

    Forked from ArkNill/browsegrab

    Token-efficient browser agent for local LLMs — Playwright + accessibility tree + MarkGrab, MCP native.

    Python

  6. diffgrab diffgrab Public

    Forked from ArkNill/diffgrab

    Web page change tracking + structured diffs — markgrab + snapgrab integration, MCP native

    Python

Repositories

Loading
Type
Select type
Language
Select language
Sort
Select order
Showing 10 of 15 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…

Morty Proxy This is a proxified and sanitized view of the page, visit original site.