autoreskill-implement-experiment

Generating manifests for reproducible experiment bundles in AutoResearch requires processing INNOVATION_PACKET and EXPERIMENT_REVIEW_PACKET inputs. This Skill automates manifest creation by consuming locked baseline-code decisions to produce the necessary configurations and scripts for your research workflow.

Implementing experiment bundles from innovation packets involves consuming INNOVATION_PACKET and EXPERIMENT_REVIEW_PACKET inputs to map datasets and generate scripts. This Skill automates applying proposed changes against a locked baseline while producing detailed patch proofs for validation.

Yes, you need Python and YAML to configure AutoResearch experiment implementations. Python is required for executing the experiment scripts, while YAML is necessary for managing the configuration files that define dataset mappings and compute backend decisions.

AutoResearch handles compute backend decisions by evaluating your experiment requirements to route execution between local GPU, AutoDL GPU, and other backends. This Skill automates the backend selection process during the experiment bundle implementation phase.

The best way to ensure proposed code changes are locked against a baseline in AutoResearch is through automated patch proofs. This Skill enforces that all proposed modifications are evaluated against a locked baseline-code decision to guarantee reproducibility and consistency.

A limitation when using YAML configurations for AutoResearch dataset mappings is the strict dependency on predefined packet structures. You must provide valid INNOVATION_PACKET and EXPERIMENT_REVIEW_PACKET inputs; without these locked baseline-code decisions, manifest generation cannot proceed.