4.2K skills · sorted by relevance
Clarify an unclear request through one question at a time, with the agent sharing its best guess for each answer. Produce a user-confirmed statement covering the desired outcome, audience, timing, success criteria, constraints, and exclusions.
Make working code easier to read and maintain while preserving its behavior. The skill identifies tangled conditions, unclear names, duplication, and unnecessary abstractions, then guides small refactoring steps checked against existing tests and project conventions.
Implement larger changes as small, working pieces. Each piece goes through testing, verification, and a focused commit, with guidance on choosing boundaries, hiding unfinished features behind switches, and keeping changes easy to roll back.
Add logs, metrics, traces that follow requests across services, and actionable alerts to production features. This skill guides signal design, request identification, response instructions, and checks that the collected data can explain failures.
Test browser interfaces through Chrome DevTools MCP, a connection that lets an agent inspect a live browser. Check screenshots, errors, network traffic, accessibility, and performance to diagnose problems and verify fixes.
Prepare a production release with checks for code quality, security, performance, accessibility, and infrastructure. Plan monitoring, gradual feature exposure, and rollback steps, then use release metrics and reliability limits to guide rollout decisions.
Define project quality standards in CONSTRAINTS.md, connect measurable targets to checks, and inspect changes for weakened safeguards. It helps teams choose thresholds and catch skipped tests, suppressed warnings, unfinished code, or lowered standards during agent development.
Investigate failing tests, broken builds, and unexpected behavior through a structured debugging process. Reproduce the failure, narrow down its source, fix the root cause, and add a regression test that catches the same problem.
Turn an unclear feature idea into a structured specification with goals, boundaries, and testable success criteria. Split broader requests into independent modules when needed, then move through planning, task breakdown, and implementation with human review.
Turn a rough idea into a concrete concept through guided discussion. Explore alternatives, compare user value and feasibility, and produce a Markdown brief covering the recommended direction, assumptions to test, minimum scope, and explicit exclusions.
Fit regression and time series models in Python, interpret coefficients, compare alternatives, and check model assumptions. Supports statistical diagnostics and forecasting, with confidence intervals and validation steps to assess uncertainty and predictive performance.
Stress-test research plans by comparing possible scenarios, such as sample losses, unavailable materials, or methods failing to transfer. Record assumptions and evidence, compare response options, and define contingency plans and observable triggers for changing course.
Work with large scientific arrays by storing data in smaller blocks and reading selected portions. Guides compression, cloud storage, format migration, and NumPy, Dask, or Xarray integration while preserving dimensions, units, and other metadata.
Build and evaluate single-agent reinforcement learning experiments, where an agent learns actions from rewards. Use Stable Baselines3 to choose algorithms, create environments, monitor training, and save models with the normalization data needed for consistent evaluation.
Build and evaluate time-to-event models with scikit-survival, including records whose events remain unobserved when follow-up ends. Keep preprocessing separate from evaluation data, compare models, predict survival probabilities, and summarize censoring-aware metrics.
Reduce data with many features into fewer dimensions using UMAP. Create 2D or 3D views, prepare inputs for clustering, compare parameter settings, and project new samples while checking whether the results remain stable and meaningful.
Work with Waypoint models to analyze microbiome samples, the communities of microbes in a sample. Convert taxonomic abundance tables, generate numerical sample representations, configure further training for prediction tasks, and evaluate models on the Compass benchmark.
Simulate concentration changes over time from biochemical models in SBML or Antimony, and compare fixed parameter conditions. Checks model units and exports CSV results plus COMBINE archives containing models and experiment settings, then replays them to compare results.
Analyze large scientific tables with Vaex, including datasets that exceed available memory. Build calculated columns, filter rows, compute statistics, visualize aggregated data, and convert files while identifying operations that may still require substantial RAM.
Explain how input features contribute to machine-learning predictions using SHAP. Select an explanation method and reference data, validate contributions against model outputs, and create charts for individual predictions or patterns across samples, including models with multiple outputs.
Build graph neural networks, models that learn from connected nodes, with PyTorch Geometric. Supports node and graph classification, link prediction, custom datasets, and training approaches for large graphs or graphs containing multiple types of nodes and connections.
Store genomic variant records in TileDB-VCF, add samples over time, and retrieve selected samples or genomic regions. Supports batch reading, allele statistics, quality checks, and VCF/BCF export, with workflows for local storage and TileDB Cloud.
Use SymPy to simplify expressions, solve equations, and calculate derivatives, integrals, and matrices with exact symbolic results. Specify assumptions, verify solutions, and turn mathematical expressions into numerical functions, source code, or LaTeX output.
Build and test SimPy models of queues, production, logistics, and other event-driven systems. Set execution limits, monitor waiting times and resource use, and analyze repeated runs with recorded random seeds for reproducible results.