P
Prime Agent
A coding agent that can refine its own harness
AgentOpen Source
An open-source, self-improving coding harness featuring a Recursive Language Model and a Continual Harness that rewrites its own planning/verification scaffolding as it runs. Claims 95.5% on ARC-AGI-3 benchmarks.
What is Prime Agent?
Prime Agent is an open-source, self-improving coding harness built around a Recursive Language Model and a Continual Harness that rewrites its own planning and verification scaffolding as it runs.
Key features
- Recursive Language Model architecture
- Continual Harness that self-edits its planning/verification scaffolding
- Claims 95.5% on ARC-AGI-3 benchmarks
- Fully open source
Who it's for
- Researchers and builders exploring self-improving agent harnesses
- Teams benchmarking coding agents against ARC-AGI-style reasoning tasks
When not to use it
The published benchmark number reflects ARC-AGI-3, not real-world codebases — treat it as a signal, not a guarantee on your own repo.
FAQ
What makes Prime Agent's harness 'self-improving'?
Its Continual Harness rewrites the agent's own planning and verification scaffolding as it runs, rather than using a fixed harness.
Open SourceAI AgentsArtificial IntelligenceAI Coding Agents
Share this launch
Embed this badge
<a href="https://orangebot.ai/product/primeagent" target="_blank" rel="noopener noreferrer"> <img src="https://orangebot.ai/api/badge/primeagent.svg" alt="Featured on OrangeBot" width="200" height="54" /> </a>