Cognitive Trace Publishing (CTP): A Method for Auditable AI-Assisted Research

This record publishes the Cognitive Trace Publishing (CTP) method and an example Cognitive Trace Package (CTP-PKG) demonstrating how AI-assisted research workflows can be archived with auditable intellectual provenance. AI systems increasingly participate in the production of research: drafting text, exploring models, generating code, and assisting in iterative reasoning. While the final manuscript captures results, the cognitive process that produced those results—prompts, iterative dialogue, branching alternatives, and intermediate reasoning—often disappears. This creates a provenance gap in AI-assisted research. Cognitive Trace Publishing (CTP) addresses this problem by pairing the manuscript with a structured cognitive trace archive. The Cognitive Trace Package (CTP-PKG) provides a reproducible container including the manuscript, replication artifacts, indexed trace segments, transformation logs, and claim-to-evidence crosswalk mappings. This Zenodo record includes: • The full manuscript describing the CTP methodology • A CTP-Lite example package illustrating the trace structure • Research transcripts documenting the development process • Package overview and trace policy documentation The archive demonstrates how Zenodo can serve as a stable substrate for versioned AI-assisted research publication. By pairing research outputs with structured provenance records, Cognitive Trace Publishing aims to make AI-assisted scholarship auditable, reproducible, and interpretable over time. Part of the Newton Research Archive. Zenodo Concept DOI: https://doi.org/10.5281/zenodo.18868360