Reproduce “From Agent Traces to Trust: A Survey of Evidence Tracing and Execution Provenance in LLM Agents” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yiqi Wang, Jiaqi Zhang, Zhangkai Wu, Taotao Cai, Zirui Liu, Zequn Sun, Mingkai Zheng, Manqing Dong and 3 more
The authors propose execution provenance, a typed graph linking an agent’s evidence, artifacts, tool calls, memory, and decisions, alongside evidence tracing, which focuses on support for claims. Their conceptual survey classifies existing benchmarks and finds that, in the curated catalog, no benchmark evaluates the complete provenance chain, and no benchmark family has strong coverage for provenance relations or recovery.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.