OSS Radar
Human-Agent-Society/reef
Infrastructure for continually self‑improving agents. Why it matters: worth a look if you build or run autonomous agents.
GitHub description: Infrastructure for continually self‑improving agents
Stars
7.6k
Last 24h
+174est.
Last 7 days
+1.2kest.
Forks
707
Who it's for
Developers and ML engineers who want agents or models to keep improving from real interaction feedback, using either model-weight training (with GPUs) or harness/prompt optimization (with only a model API endpoint).
- Continually fine-tune a model's weights from scored user or agent interactions using Slime and SGLang
- Automatically refine an agent's harness (prompts, rules, skills) from plain-language requests using a model endpoint
- Run test-time training on hard problems with a measurable objective and correctness checker
Star history (last 1 snapshot)
Only one snapshot so far. The chart fills in as OSS Radar collects daily snapshots.
* / est. Estimated: until we have enough daily snapshots for a repo, its gain is total stars ÷ days since creation. Measured numbers replace estimates automatically.
Maintainer? Add the badge to your README
Shows your live OSS Radar rank and links back here. Updates hourly.
[](https://ossradar.net/repo/Human-Agent-Society/reef)<a href="https://ossradar.net/repo/Human-Agent-Society/reef"><img src="https://ossradar.net/badge/Human-Agent-Society/reef" alt="Trending on OSS Radar"></a>About reef
Human-Agent-Society/reef is a Python project in the AI agents category, created 44 days ago. It has 7.6k GitHub stars and 707 forks, and gained +1.2k stars over the last 7 days (early estimate), ranking #15 of 104 AI agents repos tracked by OSS Radar this week.
See more trending AI agents repos or the top AI repos this week.
- Language
- Python
- Category
- AI agents
- Created
- Aug 31, 2026
- Last push
- Oct 13, 2026
- agent-infrastructure
- ai-agents
- continual-learning
- inference
- llm
- llm-training
- reinforcement-learning
- self-improving-agents
Similar rising repos
- browser-use/browser-use+1.2kest.
Agents that use the browser. Why it matters: worth a look if you build or run autonomous agents.
- bytedance/deer-flow+1.1kest.
An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and… Why it matters: worth a look if you build or run autonomous agents.
- bojieli/ai-agent-book+946est.
《深入理解 AI Agent：设计原理与工程实践》（李博杰 著）开源主仓库：全书正文、编译版 PDF 与按章配套代码. Why it matters: worth a look if you build or run autonomous agents.
- omnigent-ai/omnigent+608est.
Omnigent is an open-source AI agent framework and meta-harness: orchestrate Claude Code, Codex, Cursor, Pi, and custom agents — swap harnesses… Why it matters: worth a look if you build or run autonomous agents.
- tigerless-labs/agent-memory+571est.
Long-term memory runtime for AI agents — plain Markdown as the source of truth, local ranked retrieval, and an independent sleep-time Manage layer… Why it matters: worth a look if you build or run autonomous agents.
- kirodotdev/KiroCrew+347est.
A persistent workspace for development work that self-improves and continues beyond one session. Why it matters: worth a look if you build or run autonomous agents.