⚙️ Configuration

For the current public runtime path, LightMem2 is configured through the
OpenClaw plugin entry in:

~/.openclaw/openclaw.json

At the moment, the runtime component is still mounted under the tokenpilot
plugin entry:

{
  "plugins": {
    "entries": {
      "tokenpilot": {
        "enabled": true,
        "config": {
          "...": "..."
        }
      }
    }
  }
}

Default Runtime Mode

The current install path applies normal mode by default and sets
plugins.slots.contextEngine to layered-context.

• conservative: stabilizer on, lighter reduction preset, eviction off
• normal: stabilizer on, balanced reduction preset, eviction off
• aggressive: stabilizer on, aggressive reduction preset, eviction on with task-state estimator on

You can switch modes in-session with:

/lightmem2 mode conservative
/lightmem2 mode normal
/lightmem2 mode aggressive

Minimal Runtime Config

{
  "plugins": {
    "entries": {
      "tokenpilot": {
        "enabled": true,
        "config": {
          "enabled": true,
          "proxyBaseUrl": "https://your-openai-compatible-endpoint/v1",
          "proxyApiKey": "your_api_key",
          "modules": {
            "stabilizer": true,
            "reduction": true,
            "eviction": false
          }
        }
      }
    }
  }
}

For a first successful run, only these fields matter:

• enabled
• proxyBaseUrl
• proxyApiKey
• modules.stabilizer
• modules.reduction
• modules.eviction

Recommended starting behavior:

• keep stabilizer enabled
• keep reduction enabled
• leave eviction off until the basic runtime path is already working

If you do not want to hardcode provider credentials in openclaw.json, you can
also supply them through environment variables before starting OpenClaw:

export LIGHTMEM2_API_KEY="your_api_key"
export LIGHTMEM2_BASE_URL="https://your-openai-compatible-endpoint/v1"

If your OpenClaw home or config path is not under the default location, you can
also override:

export LIGHTMEM2_OPENCLAW_HOME="/path/to/openclaw-home"
export OPENCLAW_CONFIG_PATH="/path/to/openclaw.json"

Advanced estimator options, reduction-pass tuning, memory settings, runtime
state layout, and debugging details are intentionally kept out of the root
README.

For the full component-level configuration reference, see:

• components/tokenpilot/README.md

📄 Citation

Please cite our paper if you use EasyEdit in your work.

@inproceedings{fang2025lightmem,
  title={LightMem: Lightweight and Efficient Memory-Augmented Generation},
  author={Jizhan Fang and Xinle Deng and Haoming Xu and Ziyan Jiang and Yuqi Tang and Ziwen Xu and Shumin Deng and Yunzhi Yao and Mengru Wang and Shuofei Qiao and Huajun Chen and Ningyu Zhang},
  booktitle={The Fourteenth International Conference on Learning Representations},
  year={2026},
  url={https://openreview.net/forum?id=dyJ0GWpjJB}
}

@misc{xu2026tokenpilotcacheefficientcontextmanagement,
  title={TokenPilot: Cache-Efficient Context Management for LLM Agents},
  author={Buqiang Xu and Zirui Xue and Dianmou Chen and Chenyang Fu and Chiyu Wu and Caiying Huang and Chen Jiang and Jizhan Fang and Xinle Deng and Yijun Chen and Yunzhi Yao and Xuehai Wang and Jin Shang and Gong Yu and Ningyu Zhang},
  year={2026},
  eprint={2606.17016},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url={https://arxiv.org/abs/2606.17016},
}

🎉Contributors

We thank all the contributors to this project, more contributors are welcome!

Other Related Projects

• LLMLingua-2 — Token-level prompt compression
• SelectiveContext — Self-information-based context reduction
• Pichay — Demand paging for LLM context windows
• MemoBrain — Executive memory for long-horizon reasoning agents
• AgentSwing — Adaptive parallel context management routing for web agents
• MemOS — Memory operating system for LLM agents
• LightMem — Lightweight memory-augmented generation

🙌 We thank all the contributors to this project, and welcome further contributions from the community. We also thank the authors of the baseline methods evaluated in our experiments, including LLMLingua-2, SelectiveContext, Pichay, MemoBrain, AgentSwing, and MemOS, for making their work publicly available.