matlab-create-experiment

Name: matlab-create-experiment
Author: matlab

815stars105forks

How to install matlab-create-experiment?

matlab/matlab-agentic-toolkit/matlab-create-experiment

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-create-experiment

Installs into the current project.

Summary

Create an experiment for the MATLAB Experiment Manager app from user code, script, or problem description. TRIGGER when: user asks to create an experiment from a script/function, wants to sweep parameters or hyperparameters, asks to compare configurations, or describes a problem

Files · 1

View on GitHub

SKILL.md

1	# Create Experiment for Experiment Manager
2
3	Plans and creates experiments in MATLAB Experiment Manager by analyzing user code or problem
4	descriptions, determining the experiment type, and generating the appropriate experiment
5	function, hyperparameter table, and experiment object.
6
7	## When to Use
8	- User has a MATLAB script/function and wants to sweep parameters or hyperparameters
9	- User asks to create an experiment or compare configurations
10	- User describes a problem and wants to explore parameter space
11	- User wants to track outputs across different parameter combinations
12
13	## When NOT to Use
14	- User is debugging an existing experiment or training failure
15	- User wants to run, resume, or view results of an existing experiment
16	- User already has a working Experiment Manager experiment set up
17	- User wants to edit experiment settings (hyperparameters, strategy, description) — direct them to make changes in the Experiment Manager app. However, editing the experiment function code (.m file) is allowed.
18	- User wants Bayesian optimization or random sampling strategy — direct them to configure from the Experiment Manager app
19
20	---
21
22	## Autonomy Model
23
24	This skill follows a guidance + autonomy pattern. Decision points are tagged:
25
26	\| Tag \| Meaning \|
27	\|-----\|---------\|
28	\| `[user]` \| Always pause and ask. Non-negotiable. \|
29	\| `[auto]` \| Agent decides silently. Escalates on failure. \|
30	\| `[depends]` \| Resolved by clarity of signals — auto if clear, ask if ambiguous. \|
31
32	After analysis, tell the user how you intend to work in one plain sentence.
33	Always add: "Let me know if you'd prefer more or less control."
34
35	---
36
37	## Phase 1: Analysis
38
39	Read the user's code or problem description and classify the experiment type.
40
41	### Type Classification
42
43	Classify into one of three types using API-level and semantic signals.
44
45	IMPORTANT: If semantic signals indicate training (even without explicit DL API calls),
46	classify as training — NOT general purpose.
47
48	\| Type \| Signals \| Template class \|
49	\|------\|---------\|----------------\|
50	\| General purpose \| No DL functions, no training indicators. Optimization, simulation, data analysis. \| `'experiments.internal.experimentTemplates.MATLABFunction'` \|
51	\| Built-in training \| Uses `trainnet`/`trainNetwork`/`trainingOptions`, standard layers, datastores. OR: file name contains "train", user mentions "classification"/"transfer learning". \| `'experiments.internal.experimentTemplates.BlankTrainnetTraining'` \|
52	\| Custom training (custom loop) \| Uses `dlnetwork`/`dlfeval`/`dlgradient`/`dlarray`, manual gradients, GANs. OR: training loop with loss computation, user mentions "fine-tune"/"detector". \| `'experiments.internal.experimentTemplates.CustomTraining'` \|
53
54	### Classification Priority
55	1. Training loop structure (epoch iteration, loss tracking) → Custom training
56	2. Sets up data/layers/options without looping → Built-in training
57	3. Neither semantic nor API signals indicate training → General purpose
58
59	### Discovery Variables
60
61	\| Variable \| Tag \| Notes \|
62	\|----------\|-----\|-------\|
63	\| Experiment type \| `[depends]` \| Auto if signals are clear; ask if ambiguous \|
64	\| Parameters to sweep \| `[depends]` \| Auto-infer from hardcoded values; ask if unclear \|
65	\| Parameter values \| `[auto]` \| Suggest 2-3 alternatives around hardcoded values \|
66	\| Init function needed \| `[auto]` \| Yes if expensive ops independent of params detected \|
67	\| Function signature \| `[auto]` \| Determined by type (see REFERENCES.md) \|
68
69	Exit criteria: type classified, parameters identified, outputs known.
70
71	---
72
73	## Phase 2: Design
74
75	Generate the experiment function and prepare the full design for user review.
76
77	1. Description — Create a 2-4 line description covering: what the experiment does,
78	parameters being swept, outputs captured per trial, and evaluation criterion.
79
80	2. Parameters — Build the parameter table (Name \| Values \| Description).
81	- Hardcoded value → suggest 2-3 alternatives around it
82	- Categorical → list reasonable alternatives
83	- `[auto]` Warn if total trials > 50 (suggest reducing values)
84
85	3. Function signature — Select based on type:
86
87	\| Type \| Signature \|
88	\|------\|-----------\|
89	\| General purpose \| `[out1, out2, ...] = func(params)` \|
90	\| Built-in (Form A, targets in datastore) \| `[trainingData, net, lossFcn, options] = func(params)` \|
91	\| Built-in (Form B, targets separate) \| `[trainingData, targets, net, lossFcn, options] = func(params)` \|
92	\|

People who liked this also viewed

6 picks

Type

matlab-create-experiment

815stars105forks

View on GitHub

How to install matlab-create-experiment?

matlab/matlab-agentic-toolkit/matlab-create-experiment

$npx -y skills add matlab/matlab-agentic-toolkit --skill matlab-create-experiment

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Create Experiment for Experiment Manager
2
3	Plans and creates experiments in MATLAB Experiment Manager by analyzing user code or problem
4	descriptions, determining the experiment type, and generating the appropriate experiment
5	function, hyperparameter table, and experiment object.
6
7	## When to Use
8	- User has a MATLAB script/function and wants to sweep parameters or hyperparameters
9	- User asks to create an experiment or compare configurations
10	- User describes a problem and wants to explore parameter space
11	- User wants to track outputs across different parameter combinations
12
13	## When NOT to Use
14	- User is debugging an existing experiment or training failure
15	- User wants to run, resume, or view results of an existing experiment
16	- User already has a working Experiment Manager experiment set up
17	- User wants to edit experiment settings (hyperparameters, strategy, description) — direct them to make changes in the Experiment Manager app. However, editing the experiment function code (.m file) is allowed.
18	- User wants Bayesian optimization or random sampling strategy — direct them to configure from the Experiment Manager app
19
20	---
21
22	## Autonomy Model
23
24	This skill follows a guidance + autonomy pattern. Decision points are tagged:
25
26	\| Tag \| Meaning \|
27	\|-----\|---------\|
28	\| `[user]` \| Always pause and ask. Non-negotiable. \|
29	\| `[auto]` \| Agent decides silently. Escalates on failure. \|
30	\| `[depends]` \| Resolved by clarity of signals — auto if clear, ask if ambiguous. \|
31
32	After analysis, tell the user how you intend to work in one plain sentence.
33	Always add: "Let me know if you'd prefer more or less control."
34
35	---
36
37	## Phase 1: Analysis
38
39	Read the user's code or problem description and classify the experiment type.
40
41	### Type Classification
42
43	Classify into one of three types using API-level and semantic signals.
44
45	IMPORTANT: If semantic signals indicate training (even without explicit DL API calls),
46	classify as training — NOT general purpose.
47
48	\| Type \| Signals \| Template class \|
49	\|------\|---------\|----------------\|
50	\| General purpose \| No DL functions, no training indicators. Optimization, simulation, data analysis. \| `'experiments.internal.experimentTemplates.MATLABFunction'` \|
51	\| Built-in training \| Uses `trainnet`/`trainNetwork`/`trainingOptions`, standard layers, datastores. OR: file name contains "train", user mentions "classification"/"transfer learning". \| `'experiments.internal.experimentTemplates.BlankTrainnetTraining'` \|
52	\| Custom training (custom loop) \| Uses `dlnetwork`/`dlfeval`/`dlgradient`/`dlarray`, manual gradients, GANs. OR: training loop with loss computation, user mentions "fine-tune"/"detector". \| `'experiments.internal.experimentTemplates.CustomTraining'` \|
53
54	### Classification Priority
55	1. Training loop structure (epoch iteration, loss tracking) → Custom training
56	2. Sets up data/layers/options without looping → Built-in training
57	3. Neither semantic nor API signals indicate training → General purpose
58
59	### Discovery Variables
60
61	\| Variable \| Tag \| Notes \|
62	\|----------\|-----\|-------\|
63	\| Experiment type \| `[depends]` \| Auto if signals are clear; ask if ambiguous \|
64	\| Parameters to sweep \| `[depends]` \| Auto-infer from hardcoded values; ask if unclear \|
65	\| Parameter values \| `[auto]` \| Suggest 2-3 alternatives around hardcoded values \|
66	\| Init function needed \| `[auto]` \| Yes if expensive ops independent of params detected \|
67	\| Function signature \| `[auto]` \| Determined by type (see REFERENCES.md) \|
68
69	Exit criteria: type classified, parameters identified, outputs known.
70
71	---
72
73	## Phase 2: Design
74
75	Generate the experiment function and prepare the full design for user review.
76
77	1. Description — Create a 2-4 line description covering: what the experiment does,
78	parameters being swept, outputs captured per trial, and evaluation criterion.
79
80	2. Parameters — Build the parameter table (Name \| Values \| Description).
81	- Hardcoded value → suggest 2-3 alternatives around it
82	- Categorical → list reasonable alternatives
83	- `[auto]` Warn if total trials > 50 (suggest reducing values)
84
85	3. Function signature — Select based on type:
86
87	\| Type \| Signature \|
88	\|------\|-----------\|
89	\| General purpose \| `[out1, out2, ...] = func(params)` \|
90	\| Built-in (Form A, targets in datastore) \| `[trainingData, net, lossFcn, options] = func(params)` \|
91	\| Built-in (Form B, targets separate) \| `[trainingData, targets, net, lossFcn, options] = func(params)` \|
92	\|

People who liked this also viewed

6 picks

Type