Back to directory

Pydantic Deep Agents

Pydantic Deep Agents is both a self-hosted, open-source terminal AI assistant in the style of Claude Code and the Python framework behind it, so a single function call gives any model a filesystem, shell, planning, memory, sub-agents, sandboxed execution and MCP support.

Pydantic Deep Agents

Pydantic Deep Agents is two things in one repository: a terminal AI assistant that works as a self-hosted alternative to Claude Code, planning, editing files, running commands, searching the web, remembering across sessions, spawning subagents and connecting to MCP servers on any model a user configures; and a Python framework, built around one function, create_deep_agent, that gives a developer the same underlying harness to build a custom assistant, research agent or coding tool.

Both the assistant and the framework run on Pydantic AI, work with Claude, GPT, Gemini or local models, and are typed and MIT-licensed. A distinguishing feature called Live Run Forking splits a single agent run into isolated parallel branches, each with its own copy-on-write filesystem overlay, steering message and budget cap, which an AI judge then merges back together.

More in Coding agents

See all 793
  • The City

    github.com

    A native macOS app that shows Claude Code runs as a city, with each project as a building, each team as a floor of robots, and every animation driven by a real event from the claude CLI's stream.

  • rig

    tryrig.ai

    A small, single-binary agent runtime that points at any OpenAI-compatible model, local or hosted, reads and edits a repository's code, runs its tests and remembers what it learned, with a loop implemented in a few hundred lines.

  • Jauvex

    jauvex.reindent.com

    An Electron desktop app that puts Claude Code, Codex and Grok coding-agent sessions side by side and lets you steer them by voice, with a small model handling fast routing decisions.