prime-agent is a CLI coding and research agent that runs long-running autonomous work through a persistent Python REPL. Working context and reusable operating patterns persist in durable harness state beyond a single chat window.
Project overview
By making a persistent Python REPL the built-in model tool and storing supplemental prompts, memories, skills, and subagent specs as durable state, it targets a concrete failure mode of chat-window agents: losing useful context and reusable patterns across sessions.
Project type
AI Agent · AI Coding
Use cases
Coding & Development · Automation
Deployment
Refer to project documentation
License
License pending
Best for
Developers and researchers who need an agent for long-running coding and research work where context and reusable patterns must outlive a single chat window.
Users comfortable driving an agent from the command line in a project directory.
Key capabilities
A persistent Python REPL is the built-in model tool; file operations, shell commands, tool use, subagents, and context management happen through code.
Limitations and risks
Terminal clipboard requests cannot confirm that the local clipboard changed, so prime-agent reports them as unconfirmed.
Operation requires a subscription or API-key model provider chosen at first launch via /login; the specific model is not documented, and the cost dependency is unclear.
Getting started
Install with `curl -fsSL https://app.primeintellect.ai/prime-agent/install.sh | sh`, then run `cd /path/to/project && prime-agent` and use `/login` to choose a model provider (subscription or API key). Setup difficulty is documented as easy.
Evidence and sources
GitHub project description: A self-improving RLM agent for coding workflows and long-running autonomous tasks.
README: The **[Continual Harness](https://arxiv.org/abs/2605.09998)** stores supplemental prompts, memories, skill descriptions, and reusable subagent specifications as durable state that…
README: **Everything is programmatic:** a persistent Python REPL is the built-in model tool; file operations, shell commands, tool use, subagents, and context management happen through co…
README: Prime Agent is built for long-running work, especially for evaluations in research. These features are available in the TUI, and when run autonomously.
README: On first launch, run `/login` to choose a subscription or API-key provider. Prime Agent works in the current directory and can run commands and modify files there.