Analyze Experiment Results
Turns ML experiment result files into comparison tables, statistics and written findings.
- Stars
- 18,482
- Licence
- MIT
- Files
- 1 file
What it does
Once installed, the agent looks for experiment output files (JSON or CSV) in folders like figures/ or results/, organizes the numbers into a comparison table against a baseline, and checks things like variation across random seeds and trends across parameter sweeps. It then writes up each finding with the observation, a possible explanation, what it means, and a suggested next experiment, and can draft updates to project notes.
Great for
- Machine learning researchers who want a quick, structured read of a batch of experiment runs
- Students comparing model variants or hyperparameter sweeps and needing clean tables
- Teams who keep experiment reports and want draft finding statements to paste in
Skip it if
- Formal statistical inference: it reports mean and standard deviation and flags outliers, but does not specify a significance test, so results should be checked by someone with statistics training before publication
- Non-ML data where the metric names and sweep logic do not apply, since the workflow is written around ML experiments
Try asking
- “analyze results in results/ and compare each model against the baseline”
- “compare the learning-rate sweep outputs and tell me whether the trend is monotonic or plateaus”
- “analyze results from the three seed runs and report mean and std for accuracy”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
analyze-results.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip analyze-results.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip analyze-results.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Point the skill at a specific folder or file, since it searches figures/ and results/ by default and may miss custom output paths.
- Tell it the baseline run and the primary metric up front so the delta calculations are meaningful.
- Review the suggested next experiments and any proposed edits to project notes before accepting them, since the skill is allowed to write and edit files.
What's inside
- SKILL.md
Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.