Skip to content
@PyModel

pymodel

AI, LLM, and machine learning solutions. Think first, then code.
README.md
PyModel — AI, LLM, and machine learning solutions

Think first, then code.

Open-source tools for coding agents, UI engineering, and verifiable AI workflows.

Website · Explore all repositories · Contact

Featured

jev-judge-mcp: typed judgment tools for MCP agents. Model judges, policy decides: auto, review, or escalate.

uvx --from 'jev-judge-mcp[typesafe]' jev-judge-mcp setup     # verify and store your TypeSafe key
uvx --from 'jev-judge-mcp[typesafe]' jev-judge-mcp install   # add the server to your agents
uvx --from 'jev-judge-mcp[typesafe]' jev-judge-mcp doctor    # check the configuration, offline

Works with Claude Code, Claude Desktop, Codex, Cursor, OpenCode, Pi, omp, and Pythinker.

niblet-skill-mcp: real screen references and a design skill for coding agents that build UI. Contract, build, check.

npx skills add PyModel/niblet-skill-mcp --skill niblet -y   # the design skill, no token needed
claude mcp add niblet -- npx -y @pymodel/niblet             # optional: real screen references

The skill works alone. The server adds real screens, licensed fonts and icons, and the design system behind a picked screen.

Projects

Agent judgment and verification

  • jev-judge-mcp: verify, screen, rank, decide, and gate with typed verdicts; policy picks auto, review, or escalate
  • jev-skill: agent skill for building Jev decisions into your own app
  • code-max: requires real test evidence before an agent claims done
  • done-means-done: a task record and Stop hook that send agents back to unfinished work
  • defensive-design: proportional failure handling, idempotency, and backpressure for production code

Coding agents

  • pythinker-code: our flagship terminal agent, with editor support over ACP
  • pythinker-cli: review-first shell agent that scans and debugs before it writes
  • claude-architect: Claude Code delegates to isolated CLI agents, then verifies before you merge
  • claude-agy-mcp: Claude Code to Antigravity CLI bridge with model fallback
  • maestro: Claude plans and reviews, Codex writes the code

UI and frontend

Agent skills

Privacy

  • watermark-remover: cleaner for hidden Unicode, token watermarks, and C2PA/EXIF metadata in your files

Work with us

Found a bug or have an idea? Open an issue in the relevant repository. For AI and machine learning work, contact hello@pymodel.com or visit pymodel.com.

Pinned Loading

  1. pythinker-code pythinker-code Public

    Next-generation AI coding agent for your terminal — reads code, edits files, runs tools, and integrates with editors via ACP.

    TypeScript 29 10

  2. designer-skill designer-skill Public

    Plug-and-play MCP that gives your coding agent UI superpowers. One-line install, zero config. Claude Code, Codex, Cursor, VS Code, and more.

    JavaScript 45 5

  3. watermark-remover watermark-remover Public

    Find and remove AI provenance signals from files you own: hidden Unicode, token watermarks, visible image marks, and C2PA/EXIF/XMP metadata. Zero-dependency Python 3.10+ core; models, GPUs, and sys…

    Python 16 4

  4. react-frontend-skills react-frontend-skills Public

    Production-grade AI agent skills for React 19, Next.js 16, TypeScript 7, testing, data, UI, and frontend architecture—ready for Codex, Claude Code, Cursor, Pi, and more.

    JavaScript 27 3

  5. css-pro-tips css-pro-tips Public

    Baseline-aware modern CSS Agent Skill: small SKILL.md router + on-demand references. Claude Code, Codex, Cursor, Copilot, Gemini CLI, OpenCode, Pi, Kiro, Pythinker Code, Aider + 70 more agents.

    JavaScript 34 2

Repositories

Showing 10 of 31 repositories
  • pythinker-code Public

    Next-generation AI coding agent for your terminal — reads code, edits files, runs tools, and integrates with editors via ACP.

    PyModel/pythinker-code's past year of commit activity
    TypeScript 29 MIT 10 0 0 Updated Oct 3, 2026
  • homebrew-tap Public

    Official Homebrew tap for PyModel tools

    PyModel/homebrew-tap's past year of commit activity
    Ruby 0 1 0 0 Updated Oct 3, 2026
  • pythinker-desktop-releases Public

    Release channel for Pythinker Desktop. Auto-update metadata and installers only; source lives in PyModel/pythinker-code.

    PyModel/pythinker-desktop-releases's past year of commit activity
    0 1 0 0 Updated Oct 3, 2026
  • matt-automation Public
    PyModel/matt-automation's past year of commit activity
    JavaScript 0 1 0 0 Updated Oct 2, 2026
  • claude-architect Public

    Claude Code delegates coding to isolated CLI agents it doesn't trust (Codex, OpenCode, Pi, Pythinker), freezes what they produce, verifies it independently, and merges only what a human approves.

    PyModel/claude-architect's past year of commit activity
    TypeScript 31 MIT 2 0 10 Updated Oct 2, 2026
  • pythinker-cli Public

    Think first, then code. Where Claude Code and Codex jump straight to writing, this stays review-first: a code reviewer, security scanner, root-cause debugger, and code generator, all in one shell-native loop.

    PyModel/pythinker-cli's past year of commit activity
    Python 21 Apache-2.0 6 0 45 Updated Oct 2, 2026
  • jev-judge-mcp Public

    Typed judgment tools for MCP agents. TypeSafe's Jev model as verify, screen, find, classify, rerank, decide, compare, extract, review, gate, and score: the model judges, policy decides auto, review, or escalate.

    PyModel/jev-judge-mcp's past year of commit activity
    Python 85 MIT 4 0 1 Updated Oct 2, 2026
  • claude-agy-mcp Public

    MCP bridge that lets Claude Code delegate heavy tasks to the Antigravity CLI (agy) — purpose-built tools, model routing with fallback, and multi-turn session continuity.

    PyModel/claude-agy-mcp's past year of commit activity
    TypeScript 7 MIT 1 0 4 Updated Oct 2, 2026
  • done-means-done Public

    Resumable completion protocol for coding agents: obligations live in a durable task record, completion is computed by a validator, and a Stop hook routes execution back to unfinished work.

    PyModel/done-means-done's past year of commit activity
    Python 5 MIT 1 0 0 Updated Oct 2, 2026
  • code-max Public

    Agent skill that makes a coding agent verify before it claims done — no stubs, no fabricated test results, evidence-backed completion reports.

    PyModel/code-max's past year of commit activity
    4 MIT 2 0 0 Updated Oct 2, 2026
View all repositories

People

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