Pranav-Karra-3301
@Pranav-Karra-3301
Public Skills
mech-interp
by Pranav-Karra-3301
Mechanistic interpretability skill for understanding neural network internals. Use when: (1) Working with TransformerLens, nnsight, or SAELens (2) Analyzing attention patterns, circuits, or features (3) Training or analyzing sparse autoencoders (SAEs) (4) Activation patching, logit lens, or direct logit attribution (5) Understanding model internals ("what is this head doing?") (6) Setting up a mech interp research project Triggers: "mech interp", "mechanistic interpretability", "TransformerLens", "SAE", "sparse autoencoder", "activation patching", "logit lens", "circuits", "features", "attention heads", "residual stream", "nnsight", "nnterp", "superposition", "polysemantic", "monosemantic", "ablation", "probing", "induction head"
ml-research
by Pranav-Karra-3301
Comprehensive skill for ML/AI research experiments and finetuning. Use when: (1) Setting up new ML research project ("create ML project", "init experiment") (2) Finetuning models ("finetune LLM", "adapt pretrained model", "LoRA", "QLoRA") (3) Training from scratch ("train model", "run experiment") (4) Debugging ML issues ("model not converging", "loss exploding", "GPU OOM") (5) Setting up experiment tracking ("add W&B", "setup MLflow") (6) Optimizing GPU usage ("batch size tuning", "memory optimization") (7) Creating visualizations ("plot training curves", "confusion matrix") (8) Auditing ML code ("check reproducibility", "review experiment") Triggers: "ML", "machine learning", "deep learning", "training", "finetuning", "PyTorch", "TensorFlow", "experiment", "GPU", "CUDA", "model", "neural network", "W&B", "MLflow", "reproducibility", "learning rate", "checkpoint", "epoch"
ci-cd
by Pranav-Karra-3301
Security-hardening skill for CI/CD pipelines and pre-commit hooks. Scans codebases to detect missing security infrastructure and generates configurations. Use when: (1) Setting up CI/CD for a new project ("add CI", "setup GitHub Actions", "create workflow") (2) Adding pre-commit hooks ("add husky", "setup pre-commit", "add git hooks") (3) Security hardening ("add SAST", "setup CodeQL", "add dependabot", "security scan") (4) Auditing existing CI/CD ("check my workflows", "what security am I missing") (5) Adding commit standards ("add commitlint", "enforce conventional commits") Triggers: "CI/CD", "GitHub Actions", "pre-commit", "husky", "CodeQL", "Semgrep", "dependabot", "SAST", "security scanning", "lint-staged", "commitlint", "git hooks"
ask-me-questions
by Pranav-Karra-3301
Interactive questioning skill to help users understand vibecoded or AI-generated codebases they've lost track of. Use when: (1) User is confused about how something works in their codebase (2) User doesn't understand code that was generated with too much AI autonomy (3) User wants to find gaps, errors, or misunderstandings in implementation (4) User says things like "I don't understand this", "how does X work", "what does this do" (5) User wants to audit their understanding of AI-generated code (6) User mentions "vibecoding", "lost track", "don't know what the agent built" Triggers: "confused", "don't understand", "how does this work", "explain", "audit my understanding", "vibecoded", "what did the agent do"
documentation-done-right
by Pranav-Karra-3301
Comprehensive guide for writing excellent technical documentation. Use when: (1) Creating docs for a new or existing codebase ("document this project", "create docs") (2) Updating documentation ("update the docs", "update docs for ../other-repo") (3) Auditing docs for accuracy ("verify docs", "check if docs are outdated", "find missing docs") (4) Creating or updating llms.txt/llms-full.txt files (5) Writing API documentation (endpoints, requests, responses) (6) Any documentation-related task including READMEs, guides, and technical references Triggers: "document", "documentation", "docs", "README", "llms.txt", "API docs", "outdated docs"