Shepherd: Programmable Meta-Agents via Reversible Execution Traces

---

Quickstart | Install | Docs | Paper | Blog | Citation

Shepherd is a runtime substrate that turns an agent's entire execution into a reversible, inspectable trace. With the run itself as data, a meta-agent can checkpoint, fork, revert, and intervene on a running agent, so meta-optimization, live supervision, and recursive self-improvement become first-class operations instead of bolt-ons.

Quickstart

Coming soon.

Installation

pip install shepherd-ai

Usage example coming soon.

Documentation

Full documentation lives at https://docs.shepherd-agents.ai/.

Examples

Coming soon.

Reproducing Paper Results

The full experiment code, the four meta-agent applications and the framework-performance microbenchmarks, lives in a companion repository: shepherd-agents/shepherd-experiments. It bundles the frozen substrate snapshot used for the paper, so the numbers stay reproducible against the exact version that produced them. See also experiments/.

Citation

@misc{yu2026shepherdenablingprogrammablemetaagents,
      title={Shepherd: Enabling Programmable Meta-Agents via Reversible Agentic Execution Traces}, 
      author={Simon Yu and Derek Chong and Ananjan Nandi and Dilara Soylu and Jiuding Sun and Christopher D Manning and Weiyan Shi},
      year={2026},
      eprint={2605.10913},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2605.10913}, 
}

License

This project is licensed under the MIT License — see the LICENSE file for details.