What problem does it solve?
Enables reliable publication, reproduction, and curation of optimization experiment results as cc.experiment COBs so collaborators can verify, compare, and filter benchmark outcomes across the Radicle network.
Core Features & Use Cases
- Publish explicit experiment measurements or import autoresearch and evo session tapes to create cc.experiment COBs that replicate across peers.
- Reproduce experiments locally with safety warnings, compute direction-aware deltas, list and filter experiments by branch, author, label, or reproduction status, and manage labels and redaction for curation.
- Use Case: After running an autoresearch optimization session, push the branch to Radicle and run publish-tape to share kept improvements with reviewers who can independently reproduce the benchmark.
Quick Start
Use rad-experiment publish-tape autoresearch.jsonl --yes to import and publish kept autoresearch results to the Radicle network.