Seed: Minimal, self-modifying agent harness
Seed is a minimalist, self-modifying agent framework that allows AI agents to grow their own tools and memory structures within a local directory. It avoids complex frameworks, instead relying on a simple loop that lets the agent rewrite its own system prompts and history.
Why it matters
It offers a novel approach to AI development that prioritizes agent autonomy and persistent, self-contained growth over rigid, pre-built frameworks.
A seed agent: the smallest starting point from which an agent can grow.
There is no framework here. The entire frozen layer is seed.py — a small loop that connects a language model to exactly one tool ( exec , which runs bash) and loads its system prompt from a file the agent itself owns and may rewrite. Everything an agent normally gets from a framework — tools, memory, skills, conventions — must instead be grown by the agent, session by session, into its self/ directory.
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in