Claude Skills by thedixitjain

Use when scrutinizing statistical inference for a 《经济学(季刊)》 (China Economic Quarterly, CEQ) manuscript — choosing and justifying the clustering level, handling weak instruments with robust inference, correcting for multiple hypothesis testing, and reporting standard errors that survive a technical reviewer. Default robust SEs are rarely enough at CEQ.

Gate fine-tuned checkpoints with drift budgets, paired comparison, and forgetting checks before promotion. Use after a training run produces a checkpoint, when deciding whether a tuned model ships, or when a promoted model needs re-gating against updated goldens.

Use when packaging the artifacts behind an ACM CHI paper — prototypes, study instruments, codebooks, datasets, analysis code — for anonymous review scrutiny and for post-acceptance archival release, in a venue with no formal artifact-evaluation committee doing it for you.

Use when packaging the code, datasets, knowledge graphs, prompts, and demo systems around a CIKM paper — choosing the artifact form per track (research, applied, resource, demo), meeting the resource track's reuse-and-documentation bar, and staging anonymous review artifacts into citable public releases.

在为投向《计算机学报》(Chinese Journal of Computers, CJC) 的稿件准备代码与数据可用性材料时调用。本刊目前没有像国际计算机会议(如 ACM/USENIX artifact evaluation)那样独立的制品评审徽章制度，本技能讲清这一现状，并指导作者如何自愿地、规范地随长文提供可复现的代码、数据与实验脚本，如何在正文中声明可用性、如何托管到稳定仓库并给出访问方式，从而增强外审专家对计算机全学科实证结果的信任与可核验性。适用于让中文原创研究的支撑材料经得起三审推敲的场景。

'Stream Claude responses, use system prompts, handle multi-turn conversations, Use when working with model-inference patterns. and process structured output with the Messages API. Trigger with \"anthropic streaming\", \"claude messages api\", \"claude inference\", \"stream claude response\". '

Zero-shot image classification and image-text search.

'Tune CodeRabbit review configuration: learnings, code guidelines, and noise reduction. Use when fine-tuning review quality, training CodeRabbit with team preferences, adding code guidelines, or reducing false positives. Trigger with phrases like \"coderabbit tune reviews\", \"coderabbit learnings\", \"coderabbit guidelines\", \"reduce coderabbit noise\", \"coderabbit false positives\". '

Use when writing the response to a Cognitive Psychology (Elsevier) major/minor revision. Reviews here often demand added experiments, more model comparisons, recovery analyses, or fuller reproducibility, so the response must address every point and strengthen the model-driven inference. Structures the response letter; it does not fabricate new results or model fits.

Use when packaging the artifacts of a COLM paper — model weights, training data, prompts, evaluation sets, and cached model outputs — for anonymous review and public post-acceptance release, navigating licenses, API terms-of-service limits, and the absence of a formal COLM artifact track.

Use when deciding whether language-model research belongs at COLM or should route to ACL/EMNLP, ICLR, NeurIPS, ICML, or a workshop — applying the object-of-study test, matching against COLM's CFP lanes (training, data, evaluation, inference, safety), and weighing the trade-offs of a young venue before writing begins.

Use when deciding what evidence package a COLT (Conference on Learning Theory) paper needs, given that COLT runs no artifact-evaluation track or badges — the proof appendix is the artifact. Covers proof-verification passes, optional code companions for numerics, formalization aids, and post-acceptance release of scripts.

Use when targeting IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) or deciding whether a computer-science manuscript fits this venue. Encodes conference fit, framing, evidence bar, submission-cycle checks, rebuttal posture, and desk-reject risks for computer vision flagship.

SOTA Computer Vision Expert (2026). Specialized in YOLO26, Segment Anything 3 (SAM 3), Vision Language Models, and real-time spatial analysis.

Use when converting an accepted ACM CoNEXT paper's package into optional ACM reproducibility badges via the CoNEXT reproducibility committee — from the badge opt-in due before submission, to the one-page artifact description due within a week of acceptance, to the Available / Functional / Reusable / Reproduced criteria and evaluator-proof reuse docs.

Use when designing or auditing the evaluation of an ACM CoNEXT paper — matching evidence to claim shape with real testbeds and deployments, honest and tuned baselines, measurement statistics and uncertainty, trace and config provenance, and contamination-aware ablations for ML-for-networking work.

Use when targeting Conference and Labs of the Evaluation Forum (CLEF) or deciding whether a computer-science manuscript fits this venue. Encodes conference fit, framing, evidence bar, submission-cycle checks, rebuttal posture, and desk-reject risks for evaluation forum.

Use when targeting Conference on Health, Inference, and Learning (CHIL) or deciding whether a computer-science manuscript fits this venue. Encodes conference fit, framing, evidence bar, submission-cycle checks, rebuttal posture, and desk-reject risks for AI for health.

Use when targeting Conference on Machine Learning and Systems (MLSys) or deciding whether a computer-science manuscript fits this venue. Encodes conference fit, framing, evidence bar, submission-cycle checks, rebuttal posture, and desk-reject risks for ML systems.

'Generate confusion matrix generator operations. Auto-activating skill for ML Training. Triggers on: confusion matrix generator, confusion matrix generator Part of the ML Training skill category. Use when working with confusion matrix generator functionality. Trigger with phrases like \"confusion matrix generator\", \"confusion generator\", \"confusion\". '

Context is the complete state available to a language model at inference time. It includes everything the model can attend to when generating responses: system instructions, tool definitions, retrieved documents, message history, and tool outputs.

'Run distributed GPU training jobs on CoreWeave with multi-node PyTorch. Use when training models across multiple GPUs, setting up distributed training, or running fine-tuning jobs on CoreWeave H100 clusters. Trigger with phrases like \"coreweave training\", \"coreweave multi-gpu\", \"distributed training coreweave\", \"fine-tune on coreweave\". '

Use when packaging the artifacts of a CoRL paper — code, training configs, demonstration datasets, policy checkpoints, simulation environments, and benchmark definitions — as anonymous review-time evidence and as durable public releases after acceptance, in a venue with no formal artifact-badging track.

Use when designing or auditing experiments for a CoRL robot-learning paper — seeds and evaluation-episode counts, task-suite breadth, real-robot versus simulation evidence, sim-to-real gap measurement, baseline fairness across BC/RL/VLA families, generalization splits, and statistics for success-rate claims.

Run the corpus benchmark — booster locally, optional Gemini/Sonnet/Opus baselines — and persist a verifiable measured-vs-claimed table

Burn-rate trend over time with optional drift-alert exit code. Bins session spend into buckets, surfaces window-over-window delta, and can exit 1 when latest bucket exceeds prior mean by a configurable %. Distinct from `cost-trend` (benchmark drift); this tracks PRODUCTION spend trajectory.

'Configure cross validation setup operations. Auto-activating skill for ML Training. Triggers on: cross validation setup, cross validation setup Part of the ML Training skill category. Use when working with cross validation setup functionality. Trigger with phrases like \"cross validation setup\", \"cross setup\", \"cross\". '

Use when CrossFrame Suite routes explicit Chinese notes for books, theories, articles, excerpts, bidirectional reading, absorption, or conflict mapping.

Use when packaging what stands behind a CSCW paper — systems, analysis pipelines, codebooks, instruments, datasets from real communities — for review-time scrutiny and post-acceptance release, where community-data ethics constrain release more than any badge checklist.

在为《计算机科学》(Computer Science, JSJKX) 处理代码与数据可用性时调用。本刊是计算机全学科中文综合月刊(CCF 会刊、B 类、T2 级)，核验中未见公开的独立 artifact 徽章评审制度(待核实)，因此本技能重点讲清\"本刊现状 + 作者应做法\"：即便没有单独的 artifact 评审，如何主动提供可复现的代码与数据链接、可用性声明、校验和与许可，让单盲外审更信任你的工作，并为录用后补充材料做准备。适用于希望以开放代码与数据增强稿件可信度、又需要准确了解本刊无独立徽章制度这一现状的场景。

Use when packaging code, models, datasets, or demo videos for a CVPR paper at either review time or release time, covering anonymous supplement packaging under the external-link ban, the dataset-release-by-camera-ready rule, model-weight and license decisions, and making a vision artifact runnable by a skeptical stranger.

Use when designing or auditing the experimental program of a CVPR paper, covering benchmark and baseline selection under matched-compute fairness, the ablation study reviewers treat as mandatory, qualitative and failure-case evidence, efficiency metrics tied to the Compute Reporting Form, and generalization tests beyond a single dataset.

Use when deciding what belongs in a CVPR supplementary upload versus the 8-page body, covering the one-week-later supplement deadline, video and qualitative-result norms in computer vision, anonymous code packaging, the no-external-links rule, and keeping decision-critical evidence out of material reviewers may skip.

Use when packaging the code, benchmarks, and flows behind an ACM/IEEE Design Automation Conference (DAC) Research Manuscript into a credible, reusable artifact — given that DAC has historically run no formal artifact-evaluation or badge-issuing track (verify per cycle), so the goal is reviewer credibility and community reuse via open EDA flows and DOI-archived releases, not an ACM badge.

Use when designing or auditing the empirical evaluation of an ACM/IEEE Design Automation Conference (DAC) Research Manuscript, covering standard EDA benchmark suites (ISPD, EPFL, ISCAS/ITC, TAU, CircuitNet, OpenROAD flows), fair state-of-the-art baselines, QoR/PPA reporting with runtime, per-benchmark honesty, ablations that isolate the mechanism, and contamination-aware ML-for-EDA evaluation.

Distributed computing for larger-than-RAM pandas/NumPy workflows. Use when you need to scale existing pandas/NumPy code beyond memory or across clusters. Best for parallel file processing, distributed ML, integration with existing pandas code. For out-of-core analytics on single machine use vaex; for in-memory speed use polars.

'Process data normalization tool operations. Auto-activating skill for ML Training. Triggers on: data normalization tool, data normalization tool Part of the ML Training skill category. Use when working with data normalization tool functionality. Trigger with phrases like \"data normalization tool\", \"data tool\", \"data\". '

Expert data scientist for advanced analytics, machine learning, and statistical modeling. Handles complex data analysis, predictive modeling, and business intelligence.

Pythonic wrapper around RDKit with simplified interface and sensible defaults. Preferred for standard drug discovery including SMILES parsing, standardization, descriptors, fingerprints, clustering, 3D conformers, parallel processing. Returns native rdkit.Chem.Mol objects. For advanced control or custom parameters, use rdkit directly.

Prepare, format, and validate datasets for supervised fine-tuning and preference training. Use when converting raw data into training format, applying chat templates, configuring sequence packing, generating synthetic training data, or writing a dataset card before a run.

'Create dataset loader creator operations. Auto-activating skill for ML Training. Triggers on: dataset loader creator, dataset loader creator Part of the ML Training skill category. Use when working with dataset loader creator functionality. Trigger with phrases like \"dataset loader creator\", \"dataset creator\", \"dataset\". '

Molecular ML with diverse featurizers and pre-built datasets. Use for property prediction (ADMET, toxicity) with traditional ML or GNNs when you want extensive featurization options and MoleculeNet benchmarks. Best for quick experiments with pre-trained models, diverse molecular representations. For graph-first PyTorch workflows use torchdrug; for benchmark datasets use pytdc.

'This skill covers deploying anomaly detection systems for industrial control environments using machine learning models trained on OT network baselines, physics-based process models, and behavioral analysis of industrial protocol communications. It addresses building normal behavior profiles for SCADA polling patterns, detecting deviations in Modbus/DNP3/OPC UA traffic, identifying rogue devices, and correlating network anomalies with physical process data from historians. '

'Performs statistical analysis of Zeek conn.log connection intervals to detect C2 beaconing patterns. Uses the ZAT library to load Zeek logs into Pandas DataFrames, calculates inter-arrival time standard deviation, and flags periodic connections with low jitter. Use when hunting for command-and-control callbacks in network data. '

'Detects command-and-control (C2) communications tunneled through DNS protocol including DNS tunneling tools (Iodine, dnscat2, dns2tcp, Cobalt Strike DNS beacon), domain generation algorithms (DGA), encoded payload delivery via TXT/CNAME records, and DNS beaconing patterns. Covers Shannon entropy analysis of query subdomains, statistical anomaly detection, ML-based DGA classification, passive DNS correlation, and Zeek/Suricata signature development. Activates for requests involving DNS-based ...

'Process identify anomalies and outliers in datasets using machine learning algorithms. Use when analyzing data for unusual patterns, outliers, or unexpected deviations from normal behavior. Trigger with phrases like \"detect anomalies\", \"find outliers\", or \"identify unusual patterns\". '

'Detect anomalies in DNP3 (Distributed Network Protocol 3) communications used in SCADA systems by monitoring for unauthorized control commands, firmware update attempts, protocol violations, and deviations from baseline traffic patterns using deep packet inspection and machine learning approaches. '

Analyze git diffs for risk scoring, reviewer recommendations, and change classification. Use when preparing a PR, reviewing a large or cross-module change, or before merging to assess risk and pick reviewers.

Evaluate directories, listings, profiles, citations, associations, and resource indexes for legitimate link-building value, editorial quality, spam risk, local relevance, and submission fit. Use when deciding whether to submit to a directory, audit directory backlinks, compare listing opportunities, or reject low-quality indexes.

Pretrain LLMs at scale with PyTorch 4D parallelism.