ExecWeave: Cross-Layer Execution Provenance for Observable Agentic Software

Abstract: Agentic software invokes tools, launches processes, accesses files, and communicates with local or remote services. Application traces describe logical operations, whereas operating-system telemetry records machine activity without necessarily identifying the responsible agent action. Relating these evidence planes is a reconstruction problem: a plausible but unsupported association can assign an effect to the wrong principal. We present ExecWeave, a local-first system that materializes provider semantics and independently observed runtime events into a typed execution graph. The implementation retains separate source artifacts, distinguishes logical occurrences from repeated evidence, and records attribution, inference, and causal-status metadata. An optional command-to-process correlator emits a non-causal inferred bridge only when its bounded search yields one compatible observed process; uniqueness within this observed candidate set is not a guarantee of causal correctness. OS-observed file and network effects acquire a semantic tool or agent principal only through an observed process plus a justified command-to-process association; session-correlated file-system events do not by themselves provide that principal. Source code is available at https://github.com/Irish-kw/ExecWeave.