agentaudit
Copyright 2026 Zoltan Varga

This product includes software developed as part of the Dispatch Fidelity Benchmark,
a pre-registered measurement of dispatch fidelity in multi-agent LLM orchestrations.

  Varga, Z. "Who Validates the Validator? Instrument failure in the shape of the
  hypothesis." Zenodo. https://doi.org/10.5281/zenodo.21812041
  Pre-registration: OSF 4rgey

The scoring rules in src/agentaudit/fidelity/scorer.py, the run-binding checks in
src/agentaudit/fidelity/binding.py and the injection classes in src/agentaudit/inject/
are derived from that deposit's code and its correction protocol. The comments in those
files preserve the dated correction history, because a reader deciding whether to trust
this instrument needs to see where it has already been wrong.

Code in this repository is licensed under the Apache License, Version 2.0 (see LICENSE).
Documentation under docs/ and README.md is licensed under CC BY 4.0.
