About
Motivation
The reproducibility literature shows that process information is exactly what gets lost when results are shared. Standard task trackers and experiment trackers do not model research phases; nothing records the role of generative AI in the workflow in a machine-readable way.
In the era of AI-assisted research, there is a need for a citable, versionable record that:
- discloses the role AI played (
cognitive_prosthesis,co_pilot,autonomous_agent), and - anchors each step to verifiable evidence — commits, QPU job IDs, data/code links
so that "how the research happened" is as reproducible as the results themselves.
research-timeline provides that record with a minimal, CI-friendly, schema-validated tool.
Design philosophy
- Simplicity over power — no database, no cloud, no lock-in
- Git-native — a single JSON file that diffs cleanly and versionizes with your code
- Offline-first — works on a plane, in a lab, or in a notebook without network
- Decades-durable — plain JSON readable by any parser, archived via Zenodo and Software Heritage
How it differs from other tools
- Notes/task tools (Notion, Obsidian, Logseq, Trello) — no typed research phases, no JSON schema, no CI validation, cloud-bound
- Experiment trackers (Weights & Biases, MLflow, DVC) — track model runs and artifacts, not researcher-level process events, and no paper-oriented exports
- Notebooks (Jupyter, Quarto) — rich narrative but unstructured, no schema enforcement, no machine-readable JSON-LD export
- Lab notebooks (ELN, Code Ocean) — heavyweight, instrument-locked, or cloud-bound
research-timeline fills a specific slot: a provenance instrument for the research process. RETTIFICA (01/09/2026): an earlier claim of uniqueness was retracted - timeline visual tools exist (timeline-fishbone-generator, timeline-maker, SubThesis Timeline Generator, Research-Timeline-Planner, The Timeline Project); this tool differs by design, not by existence.
Credits & provenance
- Author: Alessandro Tulli (independent researcher, N47Lab)
- Used by: the PASM DTC Discovery program on IBM Quantum hardware (see Examples)
- Archived: Zenodo · Software Heritage
- Peer review: pyOpenSci submission #338
License
MIT — see LICENSE.