1.3K skills · sorted by relevance
Manage machine learning experiments with MLflow by recording settings, metrics, files, and trained models. Covers experiment comparison, model version management, reproducible project configurations, and deployment workflows across local and cloud environments.
Turn A/B tests and product experiment records into decision reports. Organize hypotheses, setup, and metrics, explain findings and uncertainty, then recommend shipping, iterating, extending, or stopping, with follow-up experiments based only on supplied data.
Collect experiment reports from issues and organize them in an existing Markdown index. It places entries in relevant sections, merges WandB experiment links into existing records, and checks for duplicate entries and formatting inconsistencies.
Design or revise Marin experiment pipelines by connecting reusable tasks and their outputs in a dependency graph. Organize shared implementations separately from experiment settings, then check that the workflow connects the intended inputs and stages.
How to read experiment results without fooling yourself. Confidence intervals, p-values, multiple testing, sequential testing, CUPED, heterogeneous treatment effects, ratio metrics, network effects, dashboard reconciliation, and the interpretation failures that produce confidently wrong shipping decisions. Use this skill whenever the user is reading a finished experiment result panel and about to make a ship, kill, or iterate decision, or when an experiment number does not match the dashboard number. Triggers on read experiment results, result panel, ship or kill decision, p-value, confidence interval, statistical significance, multiple testing, peeking, sequential testing, CUPED, variance reduction, heterogeneous treatment effects, ratio metric, network effects, inconclusive test, experiment versus dashboard mismatch. Use `experiment-design` instead when the test has not run yet and the question is hypothesis, sample size, duration, or what to test.
The English summary is being prepared. Showing the original description.
Experiment executor and monitor for academic research. 2-agent system covering code experiments (ML training, statistical analysis, ETL, simulation) and human studies (surveys, field studies, interviews). 4 modes: run (execute + monitor code), manage (track human studies), validate (statistical interpretation + reproducibility verification), plan (Socratic experiment design). Use when a researcher wants to run or monitor a research experiment, manage or resume a human-subject study, check the statistics or reproducibility of research results, or design an experiment or study, in English or Chinese (e.g., 跑實驗、管理研究、驗證結果、規劃實驗).
The English summary is being prepared. Showing the original description.
Guide for multi-experiment batch operations: QC screening, batch download, comparison, and report generation across many ENCODE experiments simultaneously. Use when users need to process 5+ experiments together, create experiment comparison tables, perform batch quality checks, or generate summary reports. Trigger on: batch analysis, multiple experiments, bulk processing, experiment comparison, batch QC, multi-sample, batch download, experiment table, summary report, collection analysis.
The English summary is being prepared. Showing the original description.
Autonomous growth experimentation framework: creates experiments with hypotheses, logs data points, runs statistical analysis (bootstrap CI + Mann-Whitney U), auto-promotes winners to a living playbook, and suggests next experiments. Use when asked to run growth experiments, analyze A/B tests, build experiment scorecards, or generate pacing alerts. For SEO-specific experiments see seo-ops; for outbound experiments see outbound-engine.
The English summary is being prepared. Showing the original description.
Guides experiment state transitions: launching, pausing, resuming, ending, shipping variants, archiving, resetting, duplicating, and copying to another project. Covers preconditions, implications for variant assignment and analysis, and the decision framework for when to use each action. TRIGGER when: user asks to launch, pause, resume, end, ship, archive, reset, duplicate, or copy an experiment to another project. DO NOT TRIGGER when: user is creating an experiment (use creating-experiments), configuring rollout (use configuring-experiment-rollout), or setting up metrics (use configuring-experiment-analytics).
The English summary is being prepared. Showing the original description.
Review Caveman Cloud experiment states, evaluation results, and safety checks to recommend a supported next action. Identify missing or failing evidence and explain execution blocks; the current agent interface reads experiment data without changing its lifecycle state.
Plan experiments before collecting data. Helps choose study designs, randomly assign subjects or samples, organize treatment combinations, and generate reproducible schedules while checking for confounding and repeated measurements incorrectly counted as independent samples.
Launch machine learning training jobs on local or remote GPUs. The skill checks available GPU memory, synchronizes code, saves logs, and verifies startup, with optional experiment tracking and support for Vast.ai and Modal.
Evaluate whether completed experiments support a research claim by checking cited evidence and requesting a Codex review. Identify unsupported conclusions, missing comparisons, and needed revisions, then determine whether to add experiments, change direction, or prepare for writing.
Compare machine learning experiment results from JSON or CSV files. Calculate improvements over a baseline, summarize variation across repeated runs, identify trends and unusual results, and suggest experiments to test possible explanations.
Track machine learning experiments with SwanLab by recording training settings, metrics, and media such as images and audio. Learn to compare runs, inspect training charts, and connect existing training code to local, cloud, or self-hosted dashboards.
Design controlled failure experiments for distributed systems to examine how services recover. Produce failure injection scripts, monitoring plans, rollback procedures, and exercise guides, then turn findings into documented improvements.
Design and evaluate paid-ad experiments with clear hypotheses, comparison groups, sample requirements, and stopping rules. Check data quality before estimating effects, then deliver structured results and a plain-language decision that reflects uncertainty and the population tested.
Organize experiment goals, settings, metrics, and progress in existing Obsidian notes. Keep temporary logs in daily notes, promote stable findings into result notes, and connect evidence with interpretation and decisions about the next step.
Plan experiments before collecting data by choosing how to assign samples, combine conditions, and order runs. Generate reproducible allocation schedules and experimental layouts while checking for design flaws that could make results difficult to interpret.
Try model or training changes in a copied Grug or Grugformer experiment, document the differences, and bring successful changes back into the base template while removing superseded experiment code and checking behavior.
Review GitHub experiment issues and update concise summaries that explain the setup, investigation, and current conclusion. Add useful supporting links and maintain labels so newcomers can understand both ongoing and completed work.
Preserve old experiment scripts with a dated Git tag, a named snapshot of the code, and add comments linking related GitHub issues to the archived files and cleanup pull request.
Experiment design reference for problem-class framing, Minimum Viable Experiment coaching, hypothesis formation, vetting and red flags, and experiment readiness. Use when translating a stated business outcome into candidate data-science problem classes, or when framing, vetting, scoping, or evaluating an experiment of any kind, including data feasibility, architecture, LLM, performance, use-case, UX, prototyping, and hardware experiments.
The English summary is being prepared. Showing the original description.
Create, modify, or delete experiments in Experiment Manager with live UI sync. TRIGGER when: user asks to create an experiment from code/script, wants to sweep parameters, asks to add/delete/edit experiments, or wants parameter changes reflected in the UI. Combines experiment creation (from code analysis) with frontend UI synchronization.
The English summary is being prepared. Showing the original description.