Guide users through curating the library knowledge index — reviewing block categories, marking common/important blocks, and improving block descriptions via agent inference for better agent block selection.
npx skills add https://github.com/matlab/simulink-agentic-toolkit --skill curating-library-kg
Curate .satk/library-kg/ for better agent block selection. Infer categories and descriptions from block metadata in .satk/library-cache/*.json, save curation to .satk/library-curation.json, then regenerate the KG.
building-simulink-models Gate 3.configuring-block-policybuilding-simulink-modelsbuilding-simulink-modelssatk-libraries.json or .satk/reuse-libraries.json must exist with libraries declaredmetadataQuality: "low" and "medium" blocks across ALL declared libraries. These blocks have no useful information. Write a short intent-focused description for every such block to help with agent block selection. Do not skip blocks because a file is large or because they seem unrelated to the current task."high" quality blocks are encouraged but optional.Determine mode before starting. Do not proceed without mode selection.
If .satk/library-kg/index.md already exists, start at step 1 (review existing state). If not, start at step 2.
index.md and common.md, summarize current state to user (libraries, block count, categories, common blocks)..satk/library-cache/*.json files in full. If a file exceeds read limits, read it in chunks until every block has been processed. Apply Curation Rules above — write custom descriptions for blocks, prioritizing low/medium quality. Do NOT proceed to Step 4 until descriptions exist for every low/medium quality block across ALL cache files. Save to customDescriptions.commonBlocks.categoryAssignments.library.LibraryCuration.save(projectRoot, curation), then run library.kg.Populate.run(projectRoot). Present the output summary to the user.libConfig = library.LibraryConfig.load(projectRoot);
library.LibraryCatalog.getOrCreate(libConfig, projectRoot);
Read .satk/library-cache/*.json directly. Each file:
{
"libraryName": "MotorLib",
"description": "Motor control library",
"blocks": [
{
"name": "SpeedController",
"maskType": "SpeedCtrl",
"blockType": "SubSystem",
"maskDescription": "Closed-loop speed regulation with anti-windup",
"description": "",
"pathCategory": "Controllers",
"metadataQuality": "high",
"referenceBlock": "MotorLib/Controllers/SpeedController"
}
]
}
projectRoot = prefdir();
curation = library.LibraryCuration.load(projectRoot);
curation.commonBlocks = {'Speed Controller', 'Torque Estimator'};
curation.categories = struct('name', 'motors', 'description', 'Electric motors', 'keywords', {{'motor', 'drive'}});
% Use containers.Map — supports any block name as key
curation.customDescriptions = containers.Map('KeyType', 'char', 'ValueType', 'char');
curation.customDescriptions('Unit Delay') = 'Delay signal by one sample period';
curation.customDescriptions('1-D Lookup Table') = 'Interpolate output from breakpoint-value pairs';
curation.categoryAssignments = containers.Map('KeyType', 'char', 'ValueType', 'char');
curation.categoryAssignments('DC Current Controller') = 'motor-control';
library.LibraryCuration.save(projectRoot, curation);
library.kg.Populate.run(projectRoot);
Important: Use containers.Map (not struct) for customDescriptions and categoryAssignments. Struct field names cannot contain spaces or hyphens, which most Simulink block names have.
{
"customDescriptions": [
{"block": "Unit Delay", "value": "Delay signal by one sample period"},
{"block": "1-D Lookup Table", "value": "Interpolate output from breakpoint-value pairs"}
],
"categoryAssignments": [
{"block": "DC Current Controller", "value": "motor-control"}
]
}
| Field | Type | Effect |
|-------|------|--------|
| commonBlocks | cell array of strings | Always shown in common.md regardless of quality score |
| categories | struct array with .name, .description, .keywords | Defines categories with keyword matching for assignment |
| categoryAssignments | containers.Map (blockName → categoryName) | Per-block category assignment |
| customDescriptions | containers.Map (blockName → description) | Per-block custom description (intent) |
find_system, get_param on library .slx files. Read .satk/library-cache/*.json for metadata, use library.kg.Populate.run() to generate the KG, and library.kg.Query.search() for lookups..satk/library-cache/*.json or .satk/library-kg/*.md directly — they are auto-generated.library.LibraryCuration.save() only.anges with the user before saving
----
Copyright 2026 The MathWorks, Inc.
----
Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.
Write docstrings for PyTorch functions and methods following PyTorch conventions. Use when writing or updating docstrings in PyTorch code.
Statistical models library for Python. Use when you need specific model classes (OLS, GLM, mixed models, ARIMA) with detailed diagnostics, residuals, and inference. Best for econometrics, time series, rigorous inference with coefficient tables. For guided statistical test selection with APA reporting use statistical-analysis.
Answer questions about the AI SDK and help build AI-powered features. Use when developers: (1) Ask about AI SDK functions like generateText, streamText, ToolLoopAgent, embed, or tools, (2) Want to build AI agents, chatbots, RAG systems, or text generation features, (3) Have questions about AI providers (OpenAI, Anthropic, Google, etc.), streaming, tool calling, structured output, or embeddings, (4) Use React hooks like useChat or useCompletion. Triggers on: "AI SDK", "Vercel AI SDK", "generateText", "streamText", "add AI to my app", "build an agent", "tool calling", "structured output", "useChat".
Create an llms.txt file from scratch based on repository structure following the llms.txt specification at https://llmstxt.org/
Use when working directly with the `esm` Python SDK, ESM3 or ESMC model IDs, Forge/Biohub inference clients, or ESMFold2 folding workflows.
Modal is a serverless cloud platform for running Python on demand, including on-demand GPUs. Use when deploying or serving AI/ML models, running GPU-accelerated workloads (training, fine-tuning, inference), serving web endpoints, scheduling batch jobs, or scaling Python code to cloud containers with the Modal SDK.
Use Therapeutics Data Commons through the PyTDC Python package for registry discovery, approved dataset access, task-aware splits, evaluator metrics, benchmark groups, and bounded molecular-oracle workflows.
Take matlab/curating-library-kg from the repository into ~/.claude/skills for personal
use, or into .claude/skills inside a project.
The agent identifies a skill by the name field in its header. Two skills with the
same name cannot sit side by side — one of them will be ignored.