7.1K skills · sorted by relevance
Choose a model training method and configuration from paired response preferences or individual positive and negative ratings. Covers DPO, ORPO, KTO, and SimPO, including pair construction, repeated training rounds, and troubleshooting learning settings.
Evaluate whether a fine-tuned model is ready for release by checking training data, general capability loss, comparisons with the base model, and staged production rollout. Produces an evidence-based promotion or rejection report with one prioritized remedy when rejected.
Add structured logs, performance metrics, and traces that follow requests across Python services. Establish consistent log fields and request IDs so production failures can be investigated using recorded activity and service health signals.
Evaluate companies using projected cash flows, dividends, business segments, and comparisons with peers. Select suitable valuation methods, examine how assumptions change estimated value, and check for problems that can make a seemingly cheap stock misleading.
Analyze A/B test results to decide whether to launch a change, extend testing, stop, or investigate further. Check sample size, statistical significance, uncertainty ranges, and effects on other important metrics, then summarize the recommendation and next steps.
Turn verified corrections and repeated workflows into reusable local methods. Check existing skills first, then capture when to use a method, its steps, how to validate results, and when to stop.
Turn a rough research approach into a focused plan with concrete implementation details and a small set of validation experiments. Repeated Claude reviews help refine the method while keeping the original problem and resource constraints in view.
Develop a vague research approach into a concrete method proposal through repeated Gemini reviews. Keep the original problem and constraints in focus while specifying implementation details, removing unnecessary components, and outlining the minimum experiments needed to test the claims.
Turn a vague research direction into a focused proposal and a detailed experiment roadmap. This workflow connects method refinement with validation planning, producing review records, an experiment tracker, and a summary of what to run next.
Turn a vague research approach into a concrete method proposal with focused contributions and minimal validation experiments. Ground the plan in relevant papers, then refine it through GPT-6-Astra reviews while preserving the original problem and resource constraints.
Build UI animations by first deciding whether motion serves a useful purpose. Produces code with suitable timing, transitions, interruption behavior, exit paths, reduced motion support, and hover effects adapted to the input device.
Check whether a CARLA feature is usable in the current environment and has a verified procedure. Review known failures, inspect build settings, and identify differences between client, server, and source versions before planning work.
Organize suspected trademark, copyright, patent, or trade secret infringement into a preliminary review memo. Identify factors supporting each side, missing evidence, and possible next steps, without deciding whether infringement occurred; conclusions require attorney review.
Extract PDF text using a sequence of tools when one method fails. Validate downloaded files, check whether the extracted text is substantial, record the successful method, and investigate alternative sources if extraction remains unsuccessful.
Assess enterprise Wi-Fi that uses 802.1X, a system for authenticating individual users or devices. Covers identifying authentication methods, testing rogue access points, reviewing certificate trust and storage, and assessing domain access through recovered credentials.
Review reports, spreadsheets, dashboards, or other analyses before sharing them or making decisions. Check methods, calculations, charts, and conclusions, then receive a readiness assessment with prioritized fixes, supporting evidence, and limitations that readers need to know.
Decide whether geographic data needs a map, then choose suitable background maps, visual forms, and implementation tools. Plan overlapping markers, zoom behavior, and mobile interactions so the map supports the intended analysis or product task.
Check whether Apache Hudi table maintenance is keeping pace with incoming data. Runs a read-only utility, interprets its JSON report, and explains unhealthy or skipped checks with concrete next steps.
Upgrade MSTest v3 test projects to v4 by updating packages, adapting code to changed APIs, and checking execution behavior. It also helps resolve build errors and problems finding tests in CI after an upgrade.
Check whether Codex is using Weave Router or its default provider. This skill runs a status command and summarizes the result in one line without changing any configuration.
Choose quantitative or qualitative methods that fit a research question and dataset. Compare alternatives, check assumptions, generate analysis code, interpret results, and examine sensitivity, meaning how conclusions change under different analytical choices.
Review research ethics, participant protection, and potential bias from planning through publication. The skill examines questionable research practices and study trustworthiness, then produces a structured report with evidence, risk levels, and prioritized recommendations.
Design studies that combine quantitative data, such as surveys, with qualitative evidence, such as interviews. It recommends a suitable design and outlines study phases, participant selection, timing, and ways to combine findings into an integrated interpretation.
Find datasets for a research question and compare whether they support the planned study. Receive ranked candidates with feasibility grades, coverage, access requirements, methodological limitations, and recommended next steps for obtaining the most suitable data.