[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6C1KBAZHA7CBESK5Z2CR956":3,"skill-same:01M6C1KBAZHA7CBESK5Z2CR956":223},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":36,"linkOnly":37,"markdownContent":38,"markdownBody":39,"frontMatter":40,"refName":45,"commitSha":46,"skillDirectory":47,"forksCount":13,"repositoryHtmlUrl":48,"repositoryHomepageUrl":6,"ownerHtmlUrl":49,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":50,"install":51,"resources":64},"01M6C1KBAZHA7CBESK5Z2CR956","deep-learning-book",null,"Study companion and working knowledge base for the Deep Learning textbook by Goodfellow, Bengio & Courville (MIT Press, 2016), read free at deeplearningbook.org. Indexes all 20 chapters, carries a 2016-to-2026 delta layer naming what the book got right, what was superseded (transformers, AdamW, diffusion, double descent) and what still holds, and ships four deterministic tools: a prerequisite-aware reading-path planner, a training-failure diagnostic, a capacity-and-regularization planner, and a parameter\u002FFLOP\u002Factivation-memory calculator. Use when studying or teaching this book, planning a route through it, deciding whether a chapter's advice is still current, or translating its math into a training decision. It points at the official chapters — it never reproduces them.","mne20114-hash\u002Fgest35bi","mne20114-hash","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F224168109?v=4","01M6C0JCR7334CZ39YWTZXX2NG","01M6C0JCWN0NBWMQTW1CPG2Z4G",0,"MIT","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002FSKILL.md",[],[18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35],"learning","deep learning book","deep learning","official","training","planning","engineering","Transformers","Tools","memory","knowledge","calculator","decision","planner","math","read","use","capacity","2026-11-10 10:12:30",false,"---\nname: deep-learning-book\ndescription: \"Study companion and working knowledge base for the Deep Learning textbook by Goodfellow, Bengio & Courville (MIT Press, 2016), read free at deeplearningbook.org. Indexes all 20 chapters, carries a 2016-to-2026 delta layer naming what the book got right, what was superseded (transformers, AdamW, diffusion, double descent) and what still holds, and ships four deterministic tools: a prerequisite-aware reading-path planner, a training-failure diagnostic, a capacity-and-regularization planner, and a parameter\u002FFLOP\u002Factivation-memory calculator. Use when studying or teaching this book, planning a route through it, deciding whether a chapter's advice is still current, or translating its math into a training decision. It points at the official chapters — it never reproduces them.\"\nlicense: MIT\nmetadata:\n  version: 1.0.0\n  author: Alireza Rezvani\n  category: engineering\n  updated: 2026-08-25\n---\n\n# Deep Learning — Study Companion\n\n**Source book**: *Deep Learning*, Ian Goodfellow, Yoshua Bengio & Aaron Courville\n(MIT Press, 2016) · 20 chapters, 3 parts · read free at\n[deeplearningbook.org](https:\u002F\u002Fwww.deeplearningbook.org\u002F) · companion compiled 2026-08-25.\n\n**This is a companion, not a copy.** The book is copyrighted, and its site states that the\nHTML-only format exists to discourage copying under the authors' MIT Press contract. Nothing\nhere reproduces its text. Every chapter file is original synthesis — what the chapter\nestablishes, how to use it, where it has aged — plus a link to the official chapter. Read the\nbook at the link; use this to navigate it, keep it current, and turn it into decisions.\nSee [references\u002Frights_and_use.md](references\u002Frights_and_use.md).\n\n## How to Use This Skill\n\n- **No argument** — load the core frameworks below.\n- **A topic** — ask about `regularization`, `saddle points`, `partition function`; resolved\n  through the Topic Index, then that chapter file is read before answering.\n- **`chNN`** — load that chapter's file.\n- **\"is this still true?\"** — the 2016→2026 delta layer, in every chapter file and in\n  [references\u002Fbook_to_2026_delta.md](references\u002Fbook_to_2026_delta.md).\n- **\"where do I start?\"** — run `scripts\u002Freading_path_planner.py`.\n\nWhen asked about something outside these 20 chapters, say so and route to the delta reference\nrather than improvising the book's position on material published after it.\n\n---\n\n## Core Frameworks & Mental Models\n\n### The (T, P, E) frame — ch05\n\nName the **task**, the **performance measure**, and the **experience** in one sentence before any\nmodel code. Most failed projects failed at P: an unstated metric, or a proxy whose relationship\nto the real objective was never checked.\n\n### Every loss is a negative log-likelihood — ch03, ch06\n\nChoose the output distribution, then take its negative log. Gaussian → MSE, Bernoulli → binary\ncross-entropy, categorical → cross-entropy, Laplace → MAE. \"Which loss?\" is always the question\n\"which distribution?\" in disguise. Modern contrastive and preference objectives sit outside this\nframe — a real limit of the book, not a gap in your understanding.\n\n### KL asymmetry decides your failure mode — ch03, ch19, ch20\n\nD(p‖q) ≠ D(q‖p). Forward KL is mode-covering (blurry averages); reverse KL is mode-seeking\n(sharp but partial). This single fact predicts VAE blur, GAN mode collapse, and the\ncharacteristic over-confidence of mean-field variational posteriors.\n\n### Train-error-first triage — ch11, ch05\n\nHigh training error → capacity or optimization is the bottleneck; **more data will not help**.\nLow training error with a large validation gap → data or regularization. This is the highest-value\nheuristic in the book. `scripts\u002Ftraining_diagnostics.py` runs it.\n\n### Capacity, the gap, and the U-curve's caveat — ch05, ch07\n\nRegularization trades variance for bias. But the classical U-shaped capacity curve is incomplete:\npast the interpolation threshold, test error can fall again (double descent, 2019–2020,\npost-dating the book). Practical consequence: when a large model overfits, try more data, more\nregularization or longer training **before** shrinking it.\n\n### Architecture is a prior, not a trick — ch09, ch10, ch15\n\nConvolution asserts translation equivariance and locality. Recurrence asserts that the past\ncompresses into a state. A distributed representation asserts that factors combine\ncombinatorially. When the assertion is false, the architecture cannot be rescued by tuning — and\nwhen it is true, it beats capacity. This is also why Vision Transformers need more data than\nConvNets: they discard the prior and buy it back with examples.\n\n### Depth's real cost is gradient flow and activation memory — ch06, ch08, ch10\n\nBackprop is the chain rule scheduled well: one forward-pass-equivalent of compute, and memory\nproportional to stored activations. Depth fails through vanishing\u002Fexploding gradients and\nill-conditioning, which is why residual connections, normalization and clipping exist.\n\n### The partition function organizes Part III — ch16, ch17, ch18, ch19\n\nFor undirected models, the likelihood gradient needs samples from the model itself. Four escape\nroutes: sample it (CD\u002FPCD), sidestep it algebraically (pseudolikelihood, **score matching**),\nlearn around it (NCE), or estimate it for evaluation (AIS). Score matching's descendants are\ntoday's diffusion models — which is why Part III repays reading even though its models did not\nsurvive.\n\n### Diagnose before you redesign — ch04, ch08, ch11\n\nGradient norm exploding → clip. Norm large but loss flat → ill-conditioning. Norm near zero with\nhigh loss → saturation or dead units. NaN → numerics first. Change one thing per experiment.\n\n---\n\n## Chapter Index\n\n| # | Title | Key content |\n|---|-------|-------------|\n| [ch01](chapters\u002Fch01-introduction.md) | Introduction | representation learning, depth as composition, curse of dimensionality |\n| [ch02](chapters\u002Fch02-linear-algebra.md) | Linear Algebra | norms, SVD, eigendecomposition, conditioning, PCA |\n| [ch03](chapters\u002Fch03-probability-information-theory.md) | Probability & Information Theory | distributions, entropy, KL, cross-entropy |\n| [ch04](chapters\u002Fch04-numerical-computation.md) | Numerical Computation | under\u002Foverflow, conditioning, gradient descent, KKT |\n| [ch05](chapters\u002Fch05-machine-learning-basics.md) | Machine Learning Basics | capacity, bias–variance, No Free Lunch, MLE, manifolds |\n| [ch06](chapters\u002Fch06-deep-feedforward-networks.md) | Deep Feedforward Networks | output\u002Fhidden units, universal approximation, backprop |\n| [ch07](chapters\u002Fch07-regularization.md) | Regularization | norm penalties, augmentation, early stopping, dropout |\n| [ch08](chapters\u002Fch08-optimization.md) | Optimization | SGD, momentum, init, Adam, batch norm, saddles |\n| [ch09](chapters\u002Fch09-convolutional-networks.md) | Convolutional Networks | sparse interactions, sharing, equivariance, pooling |\n| [ch10](chapters\u002Fch10-sequence-modeling.md) | Sequence Modeling | BPTT, vanishing gradients, LSTM\u002FGRU, attention |\n| [ch11](chapters\u002Fch11-practical-methodology.md) | Practical Methodology | metrics, baselines, the data-vs-capacity rule, debugging |\n| [ch12](chapters\u002Fch12-applications.md) | Applications | scaling, compression, vision, speech, NLP (dated) |\n| [ch13](chapters\u002Fch13-linear-factor-models.md) | Linear Factor Models | PPCA, factor analysis, ICA, sparse coding |\n| [ch14](chapters\u002Fch14-autoencoders.md) | Autoencoders | undercomplete, sparse, denoising, contractive |\n| [ch15](chapters\u002Fch15-representation-learning.md) | Representation Learning | transfer, distributed codes, disentanglement |\n| [ch16](chapters\u002Fch16-structured-probabilistic-models.md) | Structured Probabilistic Models | directed\u002Fundirected, energy-based, d-separation |\n| [ch17](chapters\u002Fch17-monte-carlo-methods.md) | Monte Carlo Methods | importance sampling, MCMC, Gibbs, mixing |\n| [ch18](chapters\u002Fch18-partition-function.md) | Confronting the Partition Function | CD\u002FPCD, pseudolikelihood, score matching, NCE, AIS |\n| [ch19](chapters\u002Fch19-approximate-inference.md) | Approximate Inference | ELBO, EM, mean field, amortization |\n| [ch20](chapters\u002Fch20-deep-generative-models.md) | Deep Generative Models | Boltzmann machines, VAE, GAN, autoregressive |\n\n## Topic Index\n\n- **Activation functions, ReLU, GELU** → ch06\n- **Adam, AdamW, adaptive optimizers** → ch08, ch07\n- **Attention, transformers** → ch10, ch12\n- **Autoencoders, denoising, sparse** → ch14, ch13\n- **Backpropagation, autodiff** → ch06\n- **Batch \u002F layer normalization** → ch08\n- **Bias–variance, double descent** → ch05\n- **Convolution, pooling, receptive field** → ch09\n- **Cross-entropy, KL divergence, entropy** → ch03\n- **Diffusion, score matching** → ch18, ch14, ch20\n- **Dropout, weight decay, early stopping** → ch07\n- **ELBO, variational inference, EM** → ch19\n- **Energy-based models, graphical models** → ch16\n- **GANs, VAEs, generative taxonomy** → ch20\n- **Gradient clipping, exploding\u002Fvanishing** → ch10, ch08\n- **Hyperparameter search** → ch11\n- **Initialization** → ch08\n- **LSTM, GRU, BPTT, teacher forcing** → ch10\n- **Maximum likelihood, MAP** → ch05, ch03\n- **MCMC, Gibbs, importance sampling** → ch17\n- **Numerical stability, softmax, log-space** → ch04\n- **Partition function, CD, PCD, NCE** → ch18, ch16\n- **PCA, ICA, factor analysis** → ch13, ch02\n- **Representation learning, transfer, probes** → ch15, ch01\n- **Saddle points, ill-conditioning** → ch08, ch04\n- **SVD, eigendecomposition, condition number** → ch02\n- **Training diagnostics, metric choice** → ch11\n- **Universal approximation** → ch06\n\n## Supporting Files\n\n- [glossary.md](glossary.md) — every key term with its chapter\n- [patterns.md](patterns.md) — techniques as instruments, with trade-offs\n- [cheatsheet.md](cheatsheet.md) — decision tables and defaults\n- [references\u002Fbook_to_2026_delta.md](references\u002Fbook_to_2026_delta.md) — what changed, per chapter\n- [references\u002Fprerequisite_map.md](references\u002Fprerequisite_map.md) — the real dependency graph\n- [references\u002Fstudy_method_canon.md](references\u002Fstudy_method_canon.md) — how to study a hard text\n- [references\u002Frights_and_use.md](references\u002Frights_and_use.md) — why this is a companion\n\n## Tools\n\n```bash\nS=engineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\npython3 $S\u002Freading_path_planner.py --goal \"train a transformer\" --background applied --hours-per-week 5\npython3 $S\u002Ftraining_diagnostics.py --train-loss 0.02 --val-loss 1.9 --grad-norm 0.4 --epochs 30\npython3 $S\u002Fcapacity_planner.py --params 12000000 --train-examples 50000 --train-error 0.01 --val-error 0.22\npython3 $S\u002Fmodel_arithmetic.py --spec-sample\n```\n\nEvery tool supports `--help`, `--sample` and `--output json`, uses the standard library only, and\nreturns typed exit codes.\n\n---\n\n## Scope & Limits\n\nThis companion covers the 2016 edition's 20 chapters and the delta between them and 2026\npractice. It does **not** cover: reinforcement learning beyond passing mention, LLM training\ninfrastructure, RLHF\u002FDPO alignment, agentic systems, MLOps tooling, or fairness and safety\nevaluation — none of which the book treats. For production ML engineering use\n`engineering-team\u002Fsenior-ml-engineer`; for LLM cost work use `engineering\u002Fllm-cost-optimizer`.\n\nWhen a question lands outside the book, say the book does not cover it and cite the delta\nreference for what replaced its position. A companion that quietly extrapolates is worse than one\nthat names its boundary.\n","# Deep Learning — Study Companion\n\n**Source book**: *Deep Learning*, Ian Goodfellow, Yoshua Bengio & Aaron Courville\n(MIT Press, 2016) · 20 chapters, 3 parts · read free at\n[deeplearningbook.org](https:\u002F\u002Fwww.deeplearningbook.org\u002F) · companion compiled 2026-08-25.\n\n**This is a companion, not a copy.** The book is copyrighted, and its site states that the\nHTML-only format exists to discourage copying under the authors' MIT Press contract. Nothing\nhere reproduces its text. Every chapter file is original synthesis — what the chapter\nestablishes, how to use it, where it has aged — plus a link to the official chapter. Read the\nbook at the link; use this to navigate it, keep it current, and turn it into decisions.\nSee [references\u002Frights_and_use.md](references\u002Frights_and_use.md).\n\n## How to Use This Skill\n\n- **No argument** — load the core frameworks below.\n- **A topic** — ask about `regularization`, `saddle points`, `partition function`; resolved\n  through the Topic Index, then that chapter file is read before answering.\n- **`chNN`** — load that chapter's file.\n- **\"is this still true?\"** — the 2016→2026 delta layer, in every chapter file and in\n  [references\u002Fbook_to_2026_delta.md](references\u002Fbook_to_2026_delta.md).\n- **\"where do I start?\"** — run `scripts\u002Freading_path_planner.py`.\n\nWhen asked about something outside these 20 chapters, say so and route to the delta reference\nrather than improvising the book's position on material published after it.\n\n---\n\n## Core Frameworks & Mental Models\n\n### The (T, P, E) frame — ch05\n\nName the **task**, the **performance measure**, and the **experience** in one sentence before any\nmodel code. Most failed projects failed at P: an unstated metric, or a proxy whose relationship\nto the real objective was never checked.\n\n### Every loss is a negative log-likelihood — ch03, ch06\n\nChoose the output distribution, then take its negative log. Gaussian → MSE, Bernoulli → binary\ncross-entropy, categorical → cross-entropy, Laplace → MAE. \"Which loss?\" is always the question\n\"which distribution?\" in disguise. Modern contrastive and preference objectives sit outside this\nframe — a real limit of the book, not a gap in your understanding.\n\n### KL asymmetry decides your failure mode — ch03, ch19, ch20\n\nD(p‖q) ≠ D(q‖p). Forward KL is mode-covering (blurry averages); reverse KL is mode-seeking\n(sharp but partial). This single fact predicts VAE blur, GAN mode collapse, and the\ncharacteristic over-confidence of mean-field variational posteriors.\n\n### Train-error-first triage — ch11, ch05\n\nHigh training error → capacity or optimization is the bottleneck; **more data will not help**.\nLow training error with a large validation gap → data or regularization. This is the highest-value\nheuristic in the book. `scripts\u002Ftraining_diagnostics.py` runs it.\n\n### Capacity, the gap, and the U-curve's caveat — ch05, ch07\n\nRegularization trades variance for bias. But the classical U-shaped capacity curve is incomplete:\npast the interpolation threshold, test error can fall again (double descent, 2019–2020,\npost-dating the book). Practical consequence: when a large model overfits, try more data, more\nregularization or longer training **before** shrinking it.\n\n### Architecture is a prior, not a trick — ch09, ch10, ch15\n\nConvolution asserts translation equivariance and locality. Recurrence asserts that the past\ncompresses into a state. A distributed representation asserts that factors combine\ncombinatorially. When the assertion is false, the architecture cannot be rescued by tuning — and\nwhen it is true, it beats capacity. This is also why Vision Transformers need more data than\nConvNets: they discard the prior and buy it back with examples.\n\n### Depth's real cost is gradient flow and activation memory — ch06, ch08, ch10\n\nBackprop is the chain rule scheduled well: one forward-pass-equivalent of compute, and memory\nproportional to stored activations. Depth fails through vanishing\u002Fexploding gradients and\nill-conditioning, which is why residual connections, normalization and clipping exist.\n\n### The partition function organizes Part III — ch16, ch17, ch18, ch19\n\nFor undirected models, the likelihood gradient needs samples from the model itself. Four escape\nroutes: sample it (CD\u002FPCD), sidestep it algebraically (pseudolikelihood, **score matching**),\nlearn around it (NCE), or estimate it for evaluation (AIS). Score matching's descendants are\ntoday's diffusion models — which is why Part III repays reading even though its models did not\nsurvive.\n\n### Diagnose before you redesign — ch04, ch08, ch11\n\nGradient norm exploding → clip. Norm large but loss flat → ill-conditioning. Norm near zero with\nhigh loss → saturation or dead units. NaN → numerics first. Change one thing per experiment.\n\n---\n\n## Chapter Index\n\n| # | Title | Key content |\n|---|-------|-------------|\n| [ch01](chapters\u002Fch01-introduction.md) | Introduction | representation learning, depth as composition, curse of dimensionality |\n| [ch02](chapters\u002Fch02-linear-algebra.md) | Linear Algebra | norms, SVD, eigendecomposition, conditioning, PCA |\n| [ch03](chapters\u002Fch03-probability-information-theory.md) | Probability & Information Theory | distributions, entropy, KL, cross-entropy |\n| [ch04](chapters\u002Fch04-numerical-computation.md) | Numerical Computation | under\u002Foverflow, conditioning, gradient descent, KKT |\n| [ch05](chapters\u002Fch05-machine-learning-basics.md) | Machine Learning Basics | capacity, bias–variance, No Free Lunch, MLE, manifolds |\n| [ch06](chapters\u002Fch06-deep-feedforward-networks.md) | Deep Feedforward Networks | output\u002Fhidden units, universal approximation, backprop |\n| [ch07](chapters\u002Fch07-regularization.md) | Regularization | norm penalties, augmentation, early stopping, dropout |\n| [ch08](chapters\u002Fch08-optimization.md) | Optimization | SGD, momentum, init, Adam, batch norm, saddles |\n| [ch09](chapters\u002Fch09-convolutional-networks.md) | Convolutional Networks | sparse interactions, sharing, equivariance, pooling |\n| [ch10](chapters\u002Fch10-sequence-modeling.md) | Sequence Modeling | BPTT, vanishing gradients, LSTM\u002FGRU, attention |\n| [ch11](chapters\u002Fch11-practical-methodology.md) | Practical Methodology | metrics, baselines, the data-vs-capacity rule, debugging |\n| [ch12](chapters\u002Fch12-applications.md) | Applications | scaling, compression, vision, speech, NLP (dated) |\n| [ch13](chapters\u002Fch13-linear-factor-models.md) | Linear Factor Models | PPCA, factor analysis, ICA, sparse coding |\n| [ch14](chapters\u002Fch14-autoencoders.md) | Autoencoders | undercomplete, sparse, denoising, contractive |\n| [ch15](chapters\u002Fch15-representation-learning.md) | Representation Learning | transfer, distributed codes, disentanglement |\n| [ch16](chapters\u002Fch16-structured-probabilistic-models.md) | Structured Probabilistic Models | directed\u002Fundirected, energy-based, d-separation |\n| [ch17](chapters\u002Fch17-monte-carlo-methods.md) | Monte Carlo Methods | importance sampling, MCMC, Gibbs, mixing |\n| [ch18](chapters\u002Fch18-partition-function.md) | Confronting the Partition Function | CD\u002FPCD, pseudolikelihood, score matching, NCE, AIS |\n| [ch19](chapters\u002Fch19-approximate-inference.md) | Approximate Inference | ELBO, EM, mean field, amortization |\n| [ch20](chapters\u002Fch20-deep-generative-models.md) | Deep Generative Models | Boltzmann machines, VAE, GAN, autoregressive |\n\n## Topic Index\n\n- **Activation functions, ReLU, GELU** → ch06\n- **Adam, AdamW, adaptive optimizers** → ch08, ch07\n- **Attention, transformers** → ch10, ch12\n- **Autoencoders, denoising, sparse** → ch14, ch13\n- **Backpropagation, autodiff** → ch06\n- **Batch \u002F layer normalization** → ch08\n- **Bias–variance, double descent** → ch05\n- **Convolution, pooling, receptive field** → ch09\n- **Cross-entropy, KL divergence, entropy** → ch03\n- **Diffusion, score matching** → ch18, ch14, ch20\n- **Dropout, weight decay, early stopping** → ch07\n- **ELBO, variational inference, EM** → ch19\n- **Energy-based models, graphical models** → ch16\n- **GANs, VAEs, generative taxonomy** → ch20\n- **Gradient clipping, exploding\u002Fvanishing** → ch10, ch08\n- **Hyperparameter search** → ch11\n- **Initialization** → ch08\n- **LSTM, GRU, BPTT, teacher forcing** → ch10\n- **Maximum likelihood, MAP** → ch05, ch03\n- **MCMC, Gibbs, importance sampling** → ch17\n- **Numerical stability, softmax, log-space** → ch04\n- **Partition function, CD, PCD, NCE** → ch18, ch16\n- **PCA, ICA, factor analysis** → ch13, ch02\n- **Representation learning, transfer, probes** → ch15, ch01\n- **Saddle points, ill-conditioning** → ch08, ch04\n- **SVD, eigendecomposition, condition number** → ch02\n- **Training diagnostics, metric choice** → ch11\n- **Universal approximation** → ch06\n\n## Supporting Files\n\n- [glossary.md](glossary.md) — every key term with its chapter\n- [patterns.md](patterns.md) — techniques as instruments, with trade-offs\n- [cheatsheet.md](cheatsheet.md) — decision tables and defaults\n- [references\u002Fbook_to_2026_delta.md](references\u002Fbook_to_2026_delta.md) — what changed, per chapter\n- [references\u002Fprerequisite_map.md](references\u002Fprerequisite_map.md) — the real dependency graph\n- [references\u002Fstudy_method_canon.md](references\u002Fstudy_method_canon.md) — how to study a hard text\n- [references\u002Frights_and_use.md](references\u002Frights_and_use.md) — why this is a companion\n\n## Tools\n\n```bash\nS=engineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\npython3 $S\u002Freading_path_planner.py --goal \"train a transformer\" --background applied --hours-per-week 5\npython3 $S\u002Ftraining_diagnostics.py --train-loss 0.02 --val-loss 1.9 --grad-norm 0.4 --epochs 30\npython3 $S\u002Fcapacity_planner.py --params 12000000 --train-examples 50000 --train-error 0.01 --val-error 0.22\npython3 $S\u002Fmodel_arithmetic.py --spec-sample\n```\n\nEvery tool supports `--help`, `--sample` and `--output json`, uses the standard library only, and\nreturns typed exit codes.\n\n---\n\n## Scope & Limits\n\nThis companion covers the 2016 edition's 20 chapters and the delta between them and 2026\npractice. It does **not** cover: reinforcement learning beyond passing mention, LLM training\ninfrastructure, RLHF\u002FDPO alignment, agentic systems, MLOps tooling, or fairness and safety\nevaluation — none of which the book treats. For production ML engineering use\n`engineering-team\u002Fsenior-ml-engineer`; for LLM cost work use `engineering\u002Fllm-cost-optimizer`.\n\nWhen a question lands outside the book, say the book does not cover it and cite the delta\nreference for what replaced its position. A companion that quietly extrapolates is worse than one\nthat names its boundary.\n",{"name":5,"license":14,"metadata":41,"description":7},{"author":42,"updated":43,"version":44,"category":24},"Alireza Rezvani",1787616000000,"1.0.0","main","510dd2917e051cc953dc55f56fd6ef442aa01b38","claude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi","https:\u002F\u002Fgithub.com\u002Fmne20114-hash","2026-10-28 07:50:31",{"source":8,"skillName":5,"commands":52,"prompt":62,"zipAvailable":63},[53,56,59],{"tool":54,"command":55},"npx","npx skills add mne20114-hash\u002Fgest35bi --skill deep-learning-book",{"tool":57,"command":58},"bunx","bunx skills add mne20114-hash\u002Fgest35bi --skill deep-learning-book",{"tool":60,"command":61},"pnpm","pnpm dlx skills add mne20114-hash\u002Fgest35bi --skill deep-learning-book","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002FSKILL.md",true,[65,71,77,81,85,89,93,97,101,105,109,113,117,121,125,129,133,137,141,145,149,153,156,160,164,168,172,177,181,185,189,193,197,201,205,211,215,219],{"path":66,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":69,"sourceHtmlUrl":70},"AGENTS.md","REFERENCE","参考文档",2373,"https:\u002F\u002Fgithub.com\u002Fsalexandr0s\u002Fai-config\u002Fblob\u002Fmain\u002Fshared\u002Fskills\u002Fhig-web\u002FAGENTS.md",{"path":72,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":75,"sourceHtmlUrl":76},"assets\u002Fchapter_worksheet.md","ASSET","资产文件",2348,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fassets\u002Fchapter_worksheet.md",{"path":78,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":79,"sourceHtmlUrl":80},"assets\u002Fexample_layer_spec.json",1242,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fassets\u002Fexample_layer_spec.json",{"path":82,"resourceType":73,"resourceTypeDesc":74,"sizeBytes":83,"sourceHtmlUrl":84},"assets\u002Fstudy_log_template.md",2069,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fassets\u002Fstudy_log_template.md",{"path":86,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":87,"sourceHtmlUrl":88},"chapters\u002Fch01-introduction.md",3176,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch01-introduction.md",{"path":90,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":91,"sourceHtmlUrl":92},"chapters\u002Fch02-linear-algebra.md",3174,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch02-linear-algebra.md",{"path":94,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":95,"sourceHtmlUrl":96},"chapters\u002Fch03-probability-information-theory.md",3704,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch03-probability-information-theory.md",{"path":98,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":99,"sourceHtmlUrl":100},"chapters\u002Fch04-numerical-computation.md",3104,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch04-numerical-computation.md",{"path":102,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":103,"sourceHtmlUrl":104},"chapters\u002Fch05-machine-learning-basics.md",3610,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch05-machine-learning-basics.md",{"path":106,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":107,"sourceHtmlUrl":108},"chapters\u002Fch06-deep-feedforward-networks.md",3267,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch06-deep-feedforward-networks.md",{"path":110,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":111,"sourceHtmlUrl":112},"chapters\u002Fch07-regularization.md",3644,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch07-regularization.md",{"path":114,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":115,"sourceHtmlUrl":116},"chapters\u002Fch08-optimization.md",3560,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch08-optimization.md",{"path":118,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":119,"sourceHtmlUrl":120},"chapters\u002Fch09-convolutional-networks.md",3485,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch09-convolutional-networks.md",{"path":122,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":123,"sourceHtmlUrl":124},"chapters\u002Fch10-sequence-modeling.md",3627,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch10-sequence-modeling.md",{"path":126,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":127,"sourceHtmlUrl":128},"chapters\u002Fch11-practical-methodology.md",3277,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch11-practical-methodology.md",{"path":130,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":131,"sourceHtmlUrl":132},"chapters\u002Fch12-applications.md",3122,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch12-applications.md",{"path":134,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":135,"sourceHtmlUrl":136},"chapters\u002Fch13-linear-factor-models.md",3207,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch13-linear-factor-models.md",{"path":138,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":139,"sourceHtmlUrl":140},"chapters\u002Fch14-autoencoders.md",3368,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch14-autoencoders.md",{"path":142,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":143,"sourceHtmlUrl":144},"chapters\u002Fch15-representation-learning.md",3539,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch15-representation-learning.md",{"path":146,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":147,"sourceHtmlUrl":148},"chapters\u002Fch16-structured-probabilistic-models.md",3272,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch16-structured-probabilistic-models.md",{"path":150,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":151,"sourceHtmlUrl":152},"chapters\u002Fch17-monte-carlo-methods.md",3024,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch17-monte-carlo-methods.md",{"path":154,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":115,"sourceHtmlUrl":155},"chapters\u002Fch18-partition-function.md","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch18-partition-function.md",{"path":157,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":158,"sourceHtmlUrl":159},"chapters\u002Fch19-approximate-inference.md",3363,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch19-approximate-inference.md",{"path":161,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":162,"sourceHtmlUrl":163},"chapters\u002Fch20-deep-generative-models.md",3759,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fchapters\u002Fch20-deep-generative-models.md",{"path":165,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":166,"sourceHtmlUrl":167},"cheatsheet.md",3250,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fcheatsheet.md",{"path":169,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":170,"sourceHtmlUrl":171},"glossary.md",6032,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fglossary.md",{"path":173,"resourceType":173,"resourceTypeDesc":174,"sizeBytes":175,"sourceHtmlUrl":176},"LICENSE","许可证",1056,"https:\u002F\u002Fgithub.com\u002Fsalexandr0s\u002Fai-config\u002Fblob\u002Fmain\u002Fshared\u002Fskills\u002Fhig-web\u002FLICENSE",{"path":178,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":179,"sourceHtmlUrl":180},"metadata.json",589,"https:\u002F\u002Fgithub.com\u002Fsalexandr0s\u002Fai-config\u002Fblob\u002Fmain\u002Fshared\u002Fskills\u002Fhig-web\u002Fmetadata.json",{"path":182,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":183,"sourceHtmlUrl":184},"patterns.md",4803,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fpatterns.md",{"path":186,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":187,"sourceHtmlUrl":188},"references\u002Fbook_to_2026_delta.md",6536,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Freferences\u002Fbook_to_2026_delta.md",{"path":190,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":191,"sourceHtmlUrl":192},"references\u002Fprerequisite_map.md",5007,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Freferences\u002Fprerequisite_map.md",{"path":194,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":195,"sourceHtmlUrl":196},"references\u002Frights_and_use.md",4543,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Freferences\u002Frights_and_use.md",{"path":198,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":199,"sourceHtmlUrl":200},"references\u002Fstudy_method_canon.md",5473,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Freferences\u002Fstudy_method_canon.md",{"path":202,"resourceType":67,"resourceTypeDesc":68,"sizeBytes":203,"sourceHtmlUrl":204},"rules\u002F_sections.md",16067,"https:\u002F\u002Fgithub.com\u002Fsalexandr0s\u002Fai-config\u002Fblob\u002Fmain\u002Fshared\u002Fskills\u002Fhig-web\u002Frules\u002F_sections.md",{"path":206,"resourceType":207,"resourceTypeDesc":208,"sizeBytes":209,"sourceHtmlUrl":210},"scripts\u002Fcapacity_planner.py","SCRIPT","脚本",14682,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\u002Fcapacity_planner.py",{"path":212,"resourceType":207,"resourceTypeDesc":208,"sizeBytes":213,"sourceHtmlUrl":214},"scripts\u002Fmodel_arithmetic.py",17282,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\u002Fmodel_arithmetic.py",{"path":216,"resourceType":207,"resourceTypeDesc":208,"sizeBytes":217,"sourceHtmlUrl":218},"scripts\u002Freading_path_planner.py",20125,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\u002Freading_path_planner.py",{"path":220,"resourceType":207,"resourceTypeDesc":208,"sizeBytes":221,"sourceHtmlUrl":222},"scripts\u002Ftraining_diagnostics.py",14454,"https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fengineering\u002Fdeep-learning-book\u002Fskills\u002Fdeep-learning-book\u002Fscripts\u002Ftraining_diagnostics.py",[224,261,283,306,331,351],{"skillNo":225,"name":226,"nameZh":6,"description":227,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":228,"categories":229,"keywords":243,"publishedAt":36},"01M6C27N3H0F9RGG2KASPASH04","syllabus","Generates a curated supplementary reading list from any course syllabus using Consensus academic search. Grill-me intake (syllabus input format + course audience + year range) plus a grouping forcing-options checkpoint before any search runs — so the reading list matches the course's level and recency need. Parses the syllabus to extract topics and learning outcomes, searches Consensus for recent peer-reviewed papers per topic, and produces a professionally formatted .docx with clickable Consensus links, plain-language summaries calibrated to audience level, and Bloom-higher-order discussion questions tied to course learning goals. Use when the user uploads a syllabus, course outline, or curriculum document and wants supplementary readings (e.g., 'create a reading list from this syllabus', 'find recent papers for my course') — even casual mentions with a syllabus attached should trigger this skill.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Fsyllabus\u002Fskills\u002Fsyllabus\u002FSKILL.md",[230,235,238],{"categoryNo":231,"categoryName":232,"taxonomyType":233,"taxonomyTypeDesc":234},"01JBSEEDCAT000000000000002","文档处理","SCENARIO","使用场景",{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},"01JBSEEDCAT000000000000023","Skill与智能体开发",{"categoryNo":239,"categoryName":240,"taxonomyType":241,"taxonomyTypeDesc":242},"01JBSEEDCAT000000000000034","行政与办公人员","PROFESSION","适用职业",[226,244,245,246,247,248,249,250,251,18,252,253,254,255,256,257,258,259,34,260],"docx","pr","create","intake","search","audit","integrity","checkpoint","js","format","direct","Rules","find","consensus","list","extract","outcomes",{"skillNo":262,"name":263,"nameZh":6,"description":264,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":265,"categories":266,"keywords":271,"publishedAt":36},"01M6C27DMGZ4D6ERNHSVA8H8RJ","pulse","Multi-source recency research skill that takes the pulse of any topic across Reddit, Hacker News, the open web, and optionally X\u002FTwitter within a configurable recent window (default 30 days). Forcing intake clarifies topic specificity, angle (trend\u002Fsentiment\u002Fproblems\u002Fopportunities\u002Fcomparison), time window, and platform scope before searching. Returns a synthesized briefing with citations, engagement metrics, and cross-platform pattern analysis. Use when the user requests multi-source recency intelligence on a topic (e.g., 'pulse on [topic]', 'what's happening with [topic]', 'what are people saying about [topic]', 'current conversation about [topic]', 'take the pulse of [topic]', 'trending: [topic]', 'find me info on [topic]'), and for competitor research, trend discovery, tool comparisons, and audience sentiment analysis.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Fpulse\u002Fskills\u002Fpulse\u002FSKILL.md",[267,270],{"categoryNo":268,"categoryName":269,"taxonomyType":233,"taxonomyTypeDesc":234},"01JBSEEDCAT000000000000021","资料检索与研究",{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},[263,272,273,245,274,247,249,275,276,250,277,278,279,280,254,255,281,256,282,34],"research","web","sentiment analysis","sentiment","Citations","analysis","discovery","open","tool","twitter","user",{"skillNo":284,"name":285,"nameZh":6,"description":286,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":287,"categories":288,"keywords":294,"publishedAt":36},"01M6C279M59X635VPD6Y9R616J","patent","Patent prior-art and landscape intelligence skill — not generic patent help. Commits to one of five sub-use-cases via forcing intake (novelty search \u002F freedom-to-operate \u002F competitive landscape \u002F acquisition diligence \u002F litigation prior-art) before any search runs. Searches Google Patents, Espacenet, USPTO, and optionally Lens.org for citation-graph signals. Output is an editable Word document (.docx) with verdict, ranked closest art (claim-text extracted), CPC-class-aware landscape, family-resolved hits, geographic coverage, FTO flags where applicable, strategy recommendations, and full audit log. Use when the user asks for patent searching or analysis (e.g., 'prior art search for [invention]', 'freedom to operate analysis for [product]'). Produces search signal, not legal advice — always recommends consulting a patent attorney before filing or licensing decisions. Trademark, copyright, and trade-secret questions are out of scope.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Fpatent\u002Fskills\u002Fpatent\u002FSKILL.md",[289,290,291],{"categoryNo":231,"categoryName":232,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":292,"categoryName":293,"taxonomyType":241,"taxonomyTypeDesc":242},"01JBSEEDCAT000000000000037","律师与法务",[285,244,245,295,296,297,247,248,249,277,298,299,300,255,301,302,303,304,34,305],"competitive landscape","routing","product","commits","Word","text","acquisition","decisions","strategy","coverage","advice",{"skillNo":307,"name":308,"nameZh":6,"description":309,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":310,"categories":311,"keywords":316,"publishedAt":36},"01M6C275RZWNRB3CCB14QQM7ES","notebooklm","Browser automation skill for controlling Google's NotebookLM. Use when the user wants anything done in NotebookLM (e.g., 'open NotebookLM', 'check my [name] notebook', 'ask my notebook about X', 'add [source] to NotebookLM', 'generate a Video Overview from my notebook', 'use NotebookLM Studio'). Handles reading and querying notebooks, adding sources (URLs, text, files, YouTube links, synthesized content), generating Studio outputs (Audio\u002FVideo Overviews, Mind Maps, Reports incl. Briefing Doc\u002FStudy Guide\u002FFAQ, Flashcards, Quiz, slide decks, infographics — discover the exact set from the live Studio panel; the UI evolves fast), and creating new notebooks. Requires browser automation environment — fails gracefully when unavailable.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Fnotebooklm\u002Fskills\u002Fnotebooklm\u002FSKILL.md",[312,315],{"categoryNo":313,"categoryName":314,"taxonomyType":233,"taxonomyTypeDesc":234},"01JBSEEDCAT000000000000016","图像与多媒体",{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},[308,317,318,272,319,320,321,322,323,324,325,326,279,300,254,327,328,329,330,34],"google","browser automation","automation","generate","browser","fast","SET","doc","Video","Audio","check","ask","environment","overview",{"skillNo":332,"name":333,"nameZh":6,"description":334,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":335,"categories":336,"keywords":341,"publishedAt":36},"01M6C2717B0QDJ3Y4PJP77BMW5","litreview","Academic literature orientation skill that searches papers via free keyless APIs (PubMed E-utilities + OpenAlex) by default — with the Consensus MCP as an optional enhancement lane when connected — builds a strategic search plan using PICO (default) or SPIDER \u002F Decomposition \u002F hybrid as fallbacks, and synthesizes findings into a formatted Word (.docx) research guide. Grill-me intake (research question specificity + framework hint + tentative depth) before the recon search; a second forcing checkpoint after Phase 2 confirms framework + sub-areas + depth before searches consume budget. Configurable depth (5\u002F10\u002F20 queries) controls coverage vs. speed. Output is a 'launching pad' — an orientation guide that lets a researcher dive in confidently, not a finished review. Use when the user starts literature-oriented research (e.g., 'litreview on [topic]', 'literature review on [topic]', 'I'm starting a literature review on X', 'I'm writing a paper on X', 'help me research X', 'I'm doing research on X', 'can you help me research X'). Do NOT use for single one-off paper searches wanting a quick list — that's a plain PubMed\u002FOpenAlex (or Consensus) query.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Flitreview\u002Fskills\u002Flitreview\u002FSKILL.md",[337,338,339,340],{"categoryNo":231,"categoryName":232,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":268,"categoryName":269,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":239,"categoryName":240,"taxonomyType":241,"taxonomyTypeDesc":242},[333,244,342,343,344,245,247,248,345,346,347,251,348,299,349,255,257,350,304,34],"MCP","writing","review","lane","keyless","literature review","plan","researcher","question",{"skillNo":352,"name":353,"nameZh":6,"description":354,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":355,"categories":356,"keywords":361,"publishedAt":36},"01M6C26X5G3WDVWQ40Y16ZS19F","grants","NIH grant research skill for clinical researchers. Grill-me intake (research idea + career stage + preliminary data + environment + submission posture + known institute targets) locks down the funding strategy before any search runs. Runs a 5-facet Consensus positioning analysis (with draft Significance\u002FInnovation language), maps the research to the right NIH institutes and study sections via RePORTER, finds NOSIs and funded overlap, and produces an editable Word document (.docx) with budget\u002Fscope-aware mechanism recommendations, submission timelines, and a mandatory program officer recommendation. Use when the user asks about research funding or makes any grant-related request (e.g., 'grants for [topic]', 'find grants for my research idea', 'what grants match my research', 'help me find NIH funding', 'grant opportunities for my research'). NIH-only scope — non-NIH funders (PCORI, DOD CDMRP, VA, foundations) are out of scope and flagged at intake.","https:\u002F\u002Fgithub.com\u002Fmne20114-hash\u002Fgest35bi\u002Fblob\u002Fmain\u002Fclaude-skills\u002Fresearch\u002Fgrants\u002Fskills\u002Fgrants\u002FSKILL.md",[357,358,359,360],{"categoryNo":231,"categoryName":232,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":268,"categoryName":269,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":236,"categoryName":237,"taxonomyType":233,"taxonomyTypeDesc":234},{"categoryNo":239,"categoryName":240,"taxonomyType":241,"taxonomyTypeDesc":242},[353,244,272,245,362,247,248,249,363,364,250,277,299,255,256,257,303,329,34,365],"help","data","Budget","innovation"]