4 个独立团队构建了同名为 ScienceClaw 的 AI 科研 Agent。本页帮你区分它们并选择合适的。

OpenClaw-derived (redesigned)
A self-evolving AI research colleague. 285 skills that grow at runtime — the agent writes new skills based on your usage patterns. Uses LanceDB for 4-layer persistent memory with temporal decay weighting.
Best for: Individual researchers who want an AI colleague that grows with them. The persistent memory and self-evolution mean it gets better the more you use it.
| ScienceClaw | ScienceClaw | ScienceClaw | ScienceClaw | |
|---|---|---|---|---|
| Stars | 282 | 135 | 76 | 33 |
| 架构 | OpenClaw (redesigned) | LangChain DeepAgents | Independent (Python) | OpenClaw-native |
| 技能/工具 | 285 (self-growing) | 1,900+ (ToolUniverse) | 300+ (interoperable) | 266 (markdown) |
| 部署方式 | Bash / npm | Docker (10 services) | Python venv | Bash script |
| 界面 | Web gateway | Full web app (Vue 3) | CLI + Infinite | Terminal + bots |
| 自进化 | Yes | No | Yes (mutation layer) | No |
| 持久记忆 | 4-layer LanceDB | Session-based | DAG artifacts | JSONL recall |
| 幻觉控制 | Hard protocol | Not stated | Provenance tracking | Not stated |
| 论文 | — | — | arXiv:2603.14312 | — |
| 开发方 | Individual | CAS (Beijing) | MIT LAMM Lab | Individual |
| 许可证 | MIT | MIT | MIT | MIT |