本文へ移動
cccskills
無料GitHub で公開

matlab-configure-toml-project

Use whenever "project" or "matlab.toml" is mentioned in a MATLAB context. This includes creating, editing, organizing, or configuring MATLAB projects, declaring MATLAB package dependencies or reference projects, setting up paths, lifecycle scripts, shortcuts, MATLAB packages, or asking about project structure.

インストール方法を見る

含まれるファイル(4)

  • SKILL.md12.2 KB
  • manifest.yaml475 B
  • references/deployment-reference.md13.9 KB
  • references/format-reference.md15.4 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

matlab.toml Projects

matlab.toml is a TOML-based configuration file for MATLAB projects. It defines project structure, dependencies, lifecycle scripts, shortcuts, labels, packaging, deployment, and Simulink settings. Every section is optional — the schema mandates no top-level field — but always set name, which supplies the project's display name.

matlab.toml must be placed at the project root folder — the same folder that would contain a .git/ folder or be the top-level working folder for the project.

For MATLAB R2026b and later, prefer matlab.toml for new projects. Existing .prj projects remain supported and can be converted in either direction with:

matlab.project.convertDefinitionFiles(pwd, "Toml")

Opening a project with openProject, buildtool, or runtests activates its configuration. The project's [folders].path entries then control the MATLAB path, so do not spread addpath calls through source and test files.

  • Use matlab-use-package-manager for MATLAB packages.
  • Use matlab-package-toolbox to author and build .mltbx MATLAB packages.
  • @references/format-reference.md — complete field documentation with examples, plus the validation rules that govern the whole file
  • @references/deployment-reference.md — the [compiler] deployment targets

When to Use

  • The user mentions matlab.toml, or "project" in a MATLAB context (creating, editing, or inspecting a MATLAB project's configuration).
  • Setting up MATLAB path folders, test folders, lifecycle scripts, shortcuts, package dependencies, reference projects, file labels, export profiles, MATLAB package metadata, deployment targets, or Simulink project settings.
  • Auditing an existing matlab.toml for schema violations before they cause silent failures.

When NOT to Use

  • The user wants to drive project state through the MATLAB object API directly (matlab.project.createProject, proj.addPath, etc.) rather than author or edit the file — that is a legitimate but different task; this skill's workflows assume the file is the artifact being produced.
  • The task is editing a .prj-format project definition. Do not carry .prj keys or concepts over into matlab.toml — the schemas are not equivalent. A reference project may still target a .prj project.
  • The task only concerns [compiler.*] deployment target internals in depth — read @references/deployment-reference.md directly rather than following the Creation/Editing workflows below, which are file-level, not per-target.

Key Functions

FunctionPurpose
openProjectLoads a matlab.toml (or .prj) as the current project. Does not validate the schema — unrecognized keys, missing files, and wrong types all load silently. Use it to confirm the file parses, never as a correctness check.
currentProjectReturns the currently open Project object; use its Name, PathFolders, TestFolders, StartupFiles, etc. properties to verify a round-trip after writing the file.
matlab.project.createProjectCreates a new project through the imperative API instead of authoring the file. Avoid this as the mechanism for producing matlab.toml — write the TOML directly (see Creation Workflow, step 3).

Creation Workflow

When a user wants to create a new matlab.toml (no existing file), follow this scan-first approach:

  1. Ask for the project name. This is the only required field.

  2. Scan the current folder for existing structure:

    • Find subfolders containing .m, .mlx, .slx, or .mlapp files. Propose these as [folders].path entries.
    • Identify folders whose names contain "test" (e.g., tests/, test/, unitTests/). Propose these as [folders].test entries.
    • Look for scripts matching *startup* or *setup* patterns — propose as startup-files.
    • Look for scripts matching *shutdown* or *cleanup* or *teardown* patterns — propose as shutdown-files.
    • Exclude hidden folders (.git, .svn, etc.) and common output folders (derived/, results/, build/).
  3. Generate a draft starting minimal — name, folders, and any discovered lifecycle scripts. Present it to the user.

  4. Ask about optional sections as needed, not as a checklist:

    • MATLAB packages (not MathWorks products) and reference projects
    • Shortcuts to frequently-used scripts
    • File labels and export profiles (only if used in code or explicitly requested)
    • MATLAB package metadata ([[package]]) if the project ships as an .mltbx
    • Deployment targets ([compiler.*]) if the project builds a standalone app, library, or language package
    • Simulink settings (if .slx files are present)
  5. After generating the TOML file, ask the user if they want the folders listed in [folders] to be created on disk (for nonexistent folders in new projects).

Start minimal and let the user add sections incrementally. One focused question at a time is better than a long questionnaire.

Editing Workflow

When a matlab.toml already exists and the user wants to modify it:

  1. Read the existing matlab.toml to understand the current configuration.

  2. Map the user's request to schema sections. Identify which section(s) need changes — refer to @references/format-reference.md for valid fields and formats.

  3. Make targeted edits. Modify only the sections the user asked about. Do not restructure, reformat, or reorder untouched sections. Preserve any user comments (# lines) in the file.

  4. Validate against the schema yourself. openProject does not report schema violations — unknown keys, missing files, and wrong value types all load silently and are simply ignored. A successful openProject is not evidence the file is correct, so check every key against @references/format-reference.md rather than relying on MATLAB errors and warnings:

    • Dependency values must use one of the three supported formats (package version string, package version with ID, or reference-project path).
    • Array-of-tables ([[project.profile]], [[package]], [[matlab.test.test-suite]], [[compiler.*]]) syntax must be correct.
    • No unknown keys. Every table except [user] is closed. Do not invent keys or carry over keys from .prj files.
    • Path keys must resolve. Most path values must point at files that exist and live inside the project root. Few paths (startup-folder, dependency path) can be outside the project root. Output paths (cache folders, build output) do not need to exist. The reference annotates each key. Confirming a path means listing the folder and seeing the entry — a truncated, filtered, or head-limited listing that fails to show it is not a confirmation, and neither is a ls of a guessed name that returns nothing.
    • Required fields must be present: id and package-root in [[package]], name in [[compiler.standalone]], path in project-path dependencies and shortcuts, version in version-with-ID dependencies, and both category and label in every label filter.
  5. Present the change — show the user what will be modified before writing the file.

Conventions

  • Author the TOML directly. Do not produce matlab.toml by driving matlab.project.createProject and reading back what MATLAB wrote — write the file yourself, with the sections in the order given above.
  • Tables are closed. Every table except [user] rejects unrecognized keys. Never invent a key or carry one over from a .prj file; check every key against @references/format-reference.md.
  • openProject is not a validator. A file that opens cleanly may still contain silently-ignored schema violations — the only way to be sure a key is correct is to look it up.
  • Path values must be confirmed, not guessed. Listing a folder and seeing the entry is a confirmation; a truncated, filtered, or head-limited listing that fails to show it is not, and neither is an ls of a guessed name that returns nothing.
  • MathWorks products are never [dependencies] entries. [dependencies] accepts any key, so a product name loads silently and wrongly. Record product requirements elsewhere (e.g. [project].summary, [[package]].product-dependencies, or a comment).
  • Preserve what the user didn't ask you to touch. Comments and section order in an existing file survive an edit unless the user's request required changing them.

Quick Reference

SectionPurposeKey Fields
nameProject name (always set it)string
[user]Free-form custom dataany keys/nested tables
[folders]MATLAB path and test folderspath, test (arrays)
[dependencies]Declare MATLAB packages (not MathWorks products) or reference projectspackages: version string or {version, id}; references: {path, absolute}
[project]Lifecycle and metadatasummary, startup-folder, startup-files, shutdown-files, custom-tasks
[project.dependency-analyzer]Dependency analysis cachecache-file
[project.shortcuts]Named script shortcuts{path, icon} per shortcut; can be grouped
[project.labels.Category]File categorizationLabelName = ["file.m", "dir/"]
[[project.profile]]Export/filter profilesname, included-files, excluded-files, excluded-labels
[project.dashboards]Dashboard label groupingscomponent, unit (label references)
[project.digital-thread]Traceability settingstool-output-tracking, ignore
[[package]]Build a MATLAB packageid (required), package-root (required), name, version, summary, description, folders, apps, provider
[compiler.*]Deployment build targets10 targets: standalone (name required), web-app, excel-add-in, dot-net-assembly, c-shared-library, cpp-shared-library, production-server-archive, com-component, java-package, python-package — see @references/deployment-reference.md
[matlab.test]Test runner configurationcoverage-setting, test-suite
[simulink]Simulink settingsstart-simulink, cache-folder, codegen-folder, refresh-customizations
[simulink.process-advisor]Process Advisor settingsincremental-build, enable-model-caching, 9 more string properties

[[package]] can include MATLAB code, Simulink models, apps, documentation, and other project content in the resulting MATLAB package. Use matlab-package-toolbox for the authoring and build workflow, or matlab-use-package-manager to install a completed package.

Package Dependencies and Reference Projects

Each key under [dependencies] names either a MATLAB package or a reference project. MathWorks products are not packages and must not be listed here.

Package dependency: version string — for a MATLAB package:

mytoolbox = "3.*"

Package dependency: version with ID — when the package ID is available:

motor = {version = "*", id = "9b9e5bc2-a583-4d79-98c5-d2b9af135697"}

Reference project — for a project at a filesystem path (absolute is optional and defaults to relative):

shared-utils = {path = "../shared_utils_lib"}

The referenced project can use either a legacy .prj definition or a matlab.toml definition. The key is a display label chosen by the user and is unrelated to the folder on disk. Confirm path by listing the candidate parent folder — never derive it from the dependency name or from the other project's name.

Examples

Minimal

name = "My Project"

Typical

name = "Signal Analyzer"

[folders]
path = [
    "src",
    "utilities",
]
test = [
    "tests",
]

[dependencies]
signal-processing = "2.*"

[project]
startup-files = [
    "setup/initialize.m",
]

[project.shortcuts]
"Run Analysis" = {path = "scripts/analyze.m"}

Copyright 2026 The MathWorks, Inc.


レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Guide for accessing financial and economic data in MATLAB using the Datafeed Toolbox. Covers Bloomberg (market data via bloomberg/blp/bloombergHypermedia), FRED (Federal Reserve economic data via fredrs), Haver Analytics (economic data via haver/haverdirect/haverview), and LSEG Datastream (historical data via datastreamws). Use when connecting to any of these data providers from MATLAB.

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

Read BEFORE writing any code that adds Additive White Gaussian Noise (AWGN) to signals and converts between SNR, Eb/No, Es/No, and per-subcarrier SNR for communications simulations, using awgn(), convertSNR(), berawgn(). The default MATLAB patterns for AWGN (e.g., 'measured' option, manual SNR formulas) produce subtly incorrect results. This skill specifies the correct calling conventions, required function usage, and critical anti-patterns that must be avoided.

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

Analyze AMS waveform data using Mixed-Signal Blockset utilities: phase noise measurement, clock jitter, anti-aliased resampling, timing measurements, lock time, INL/DNL, ADC/DAC calibration, HSpice import. Use when analyzing time-domain voltage from PLL/VCO/clock simulations, measuring phase noise from variable-step solver output, computing jitter, or resampling non-uniform data.

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

Design and analyze electrically large antenna structures using MATLAB Antenna Toolbox. Covers reflector antennas (parabolic, Cassegrain, Gregorian, offset, corner, cylindrical, spherical, custom STL), reflectarrays and reconfigurable intelligent surfaces (RIS), antennas installed on platforms (vehicles, aircraft, ships, satellites), and radar cross section (RCS) analysis. Includes solver selection (MoM-PO, PO, MoM, FMM), mesh control, and GPU acceleration. Use when the user wants to design a dish/reflector antenna, reflectarray, analyze an antenna on a platform, or compute RCS.

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

Analyze data using MATLAB. Use when the task involves tables, timetables, time-series data, numeric arrays, sensor matrices, or gridded data — including but not limited to exploring, row filtering, sorting, cleaning, transforming, aggregating, smoothing, padding, trimming, and answering questions about data. MATLAB provides extensive, easy-to-use built-in functions for these workflows with no additional products required.

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

S-parameters, insertion loss, fields, currents, mesh control, and solver selection for RF PCB performance validation. TRIGGER: user asks to compute S-parameters, analyze insertion/return loss, extract fields or currents, compare MoM vs FEM, or control mesh for any RF PCB component. Invoke BEFORE writing sparameters() or solver code — API is non-obvious. SKIP: designing or creating components (use the specific matlab-design-pcb-* skill), material/stackup setup only (use matlab-manage-pcb-material), optimization sweeps (use matlab-optimize-pcb-design), PDN/IR-drop analysis (use matlab-analyze-pcb-pdn).

日本語の概要は準備中です。原文の説明を表示しています。

matlab/matlab-agentic-toolkit1,1512026年10月9日 更新

matlab のスキルをすべて見る

このスキルの問題を報告する