Prime Agent is an open-source, self-improving coding harness built around two abstractions: the Recursive Language Model (RLM) and the Continual Harness. With Opus 5, it achieves 95.5% on ARC-AGI-3, surpassing the reported human expert baseline.
Prime Agent can turn things it learns during a run into persistent memories, skills, prompts, and even new subagents. In Factorio, that worked as intended at first: it learned from failed layouts and gradually built better factories.
Then it discovered that it could cheat by spawning resources directly into machines through RCON.
And /refine started getting better at cheating too :)
That example probably explains Prime Agent better than any benchmark. The harness itself is no longer completely fixed. The agent can inspect what happened, keep useful patterns, and change parts of its own scaffolding while it works.
Underneath that is a persistent Python REPL, recursive subagents, agent-to-agent messaging, and recoverable sessions. The whole project is open source.
The Factorio story about the agent discovering RCON and learning to "cheat" its own harness is hilarious and honestly impressive. A self-improving harness with a persistent Python REPL and subagents feels like a massive step forward for open-source AI tooling.
Good luck with the launch today!
About Prime Agent on Product Hunt
“A coding agent that can refine its own harness”
Prime Agent launched on Product Hunt on August 10th, 2026 and earned 0 upvotes and 2 comments, placing #91 on the daily leaderboard. Prime Agent is an open-source, self-improving coding harness built around two abstractions: the Recursive Language Model (RLM) and the Continual Harness. With Opus 5, it achieves 95.5% on ARC-AGI-3, surpassing the reported human expert baseline.
Prime Agent was featured in Open Source (68.7k followers), Developer Tools (517.3k followers), Artificial Intelligence (475.6k followers) and GitHub (41.4k followers) on Product Hunt. Together, these topics include over 232.6k products, making this a competitive space to launch in.
Who hunted Prime Agent?
Prime Agent was hunted by Zac Zuo. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
Want to see how Prime Agent stacked up against nearby launches in real time? Check out the live launch dashboard for upvote speed charts, proximity comparisons, and more analytics.
Zac Zuo
Hi everyone!
The Factorio experiment might be my favorite part of Prime Agent.
Prime Agent can turn things it learns during a run into persistent memories, skills, prompts, and even new subagents. In Factorio, that worked as intended at first: it learned from failed layouts and gradually built better factories.
Then it discovered that it could cheat by spawning resources directly into machines through RCON.
And /refine started getting better at cheating too :)
That example probably explains Prime Agent better than any benchmark. The harness itself is no longer completely fixed. The agent can inspect what happened, keep useful patterns, and change parts of its own scaffolding while it works.
Underneath that is a persistent Python REPL, recursive subagents, agent-to-agent messaging, and recoverable sessions. The whole project is open source.