PrimeIntellect-ai/prime-agent

This page introduces an external open-source repository. It is not an HDATF product.

What it does

An open-source coding and research agent that treats context as variables and calls subagents from a persistent Python REPL. It keeps prompts, memories, skills and subagent specs as durable state that it can refine and roll back.

How it helps ATF

Harness keeps goal, state and memory and checks results with evidence, so Prime Agent's evidence-backed refinement of harness state with rollback snapshots is worth comparing.

License

MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.

More in this category

• huangruiteng/loopx
  LoopX is a local-first control plane that runs on top of agent harnesses such as Codex, Claude Code and Cursor. It keeps objectives, gates, todos, evidence, quota and handoffs across turns, while the harness executes bounded work.
• ifixai-ai/iFixAi
  A tool for auditing AI agents, run through a CLI or by the agent itself as a plugin or skill. It runs inspections that mix adversarial testing with operational checks, and returns a letter grade with a scorecard.
• cloudflare/computer
  A preview TypeScript package that gives agents a virtual filesystem stored in SQLite inside a Durable Object, with pluggable execution backends: a sandbox container with a FUSE mount, a shell in a Dynamic Worker, or JavaScript modules.
• cloudflare/cloudflare-os
  An open-source AI productivity environment on Cloudflare Workers. It combines an agent chat preloaded with knowledge of how the company operates, sandboxed building and sharing of small apps, and a guardrail framework called Gatekeepers for agents and apps.
• trycompai/crm
  Comp AI CRM is an open-source CRM run by an agent that works through its own queue and schedule. Its tools report observed evidence; strong evidence is written to records, and weak evidence becomes a suggestion for a person.

Only repositories in the ranked Trendshift lists are included, and the lists are used only to find candidates. We do not copy their ranks. Descriptions, licenses and star counts come from each GitHub repository. The notes are our own reading. We have not tested these projects, and a place on a trending list does not prove quality.