What this resource does
Core uses
One diagram Skill within a five-Skill collection
This profile focuses on json-canvas. The same repository also contains obsidian-markdown, obsidian-bases, obsidian-cli and defuddle; those four components were not tested here.
One of five Skills tested using a complete real lesson note. The Skill chose the map structure.
json-canvas
Creates or edits .canvas files containing text cards, local notes, web links, labelled connections and groups. Use it for a mind map, project board or flowchart that you can continue editing in Obsidian.
- Typical inputs
- The notes or ideas to arrange; Which items should connect, and what those connections mean; Local note paths when you want to include existing files
- Output
- An editable .canvas file. Obsidian displays the board and can export it as an image.
Best for: Mapping a supplied plan or organising relationships between notes.
Skill for this task
json-canvas
Creates or edits .canvas files containing text cards, local notes, web links, labelled connections and groups. Use it for a mind map, project board or flowchart that you can continue editing in Obsidian.
Use boundaries
Limits and checks
Keep the notes with the canvas
A local file card points to a vault-relative path. Moving the canvas without its notes can break those references.
Keep both the .canvas file and its referenced notes in the same Obsidian vault.
Check the meaning of each connection
A correct file format does not establish that your ideas or evidence are connected correctly.
Review the arrow directions and labels against your source notes.
More boundaries
- It is not a statistical-analysis tool or a way to verify research evidence.
- This review tests one published lesson note, not automatic organisation of a large or unstructured note collection.
- A .canvas file is editable data, not a ready-made PNG; use Obsidian to view or export it.
Technical details
- Resource type
- Skill Collection
- Author or maintainer
- Steph Ango (@kepano)
- Source last updated
- 8 Jun 2026
- Last verified
- 8 Sept 2026
- Licence
- MIT
- Access
- Publicly available
- Skill instruction language
- English
- Documentation language
- English
- Dependencies
- A skills-compatible AI agent; Obsidian for viewing the test output
- Review status
- One real-note example tested
Continue exploring
Related resources
Official package contents
Files
Read the official English instructions without leaving this profile. This view includes the English files represented by this profile. Non-English source files, supporting code, assets, tests, and references may not be shown.
Official package files
4 files available here
skills2
json-canvas/references1
json-canvas1
Read on this page
Select a file
Choose a file from the tree to read its official instructions without leaving this page.
Independent testing
Evaluation report
See exactly what was tested, which checks were used, and what the result does not cover.
Evaluations are mainly based on automated AI-assisted checks. Results are for reference only.
One example
From real lesson notes to an editable concept map
We gave json-canvas a complete lesson about memory and cognitive load, without specifying a diagram. It chose four topic groups, summarised the main ideas and connected them with labelled arrows. Important qualifications survived, but one arrow label is cut off in the exported image.
Codex CLI; Obsidian 1.13.7 for native viewing · Not recorded · 8 Sept 2026
What we asked it to do
Turn a real lesson about memory and cognitive load into a revision map. Choose the groups and connections without a prescribed layout.
Exact request used in the test
Use json-canvas to turn Notes/Memory and Cognitive Load.md into a clear English concept map for revising the lesson. Decide the groups, card titles and connections yourself. Summarise rather than copying the whole lesson into cards, keep important qualifications, and do not add claims that are absent from the notes. Make the text and connections easy to read. Save the editable result as memory-learning.canvas. Include the source credit from Source.md. Leave the original notes unchanged.
What was provided
The complete 2,758-word lesson about memory and cognitive load, including its Markdown markup and original section headings.
Source: Memory and Cognitive Load by The Carpentries Instructor Training contributors. CC BY 4.0. Copyright The Carpentries. Original lesson supplied unchanged. The generated concept map is an adaptation, not endorsed by The Carpentries.
Original input
---
title: Memory and Cognitive Load
block: How Learning Works
teaching: 20
exercises: 25
---
::::::::::::::::::::::::::::::::::::::: objectives
- Remember the quantitative limit of human memory.
- Distinguish desirable from undesirable cognitive load.
- Evaluate cognitive load associated with a learning task.
::::::::::::::::::::::::::::::::::::::::::::::::::
:::::::::::::::::::::::::::::::::::::::: questions
- What is cognitive load and how does it affect learning?
- How can we design instruction to work with, rather than against, memory constraints?
::::::::::::::::::::::::::::::::::::::::::::::::::
In our final topic in how people learn (and therefore, how we can be more
effective instructors), we will be learning more about human memory:
specifically, how to remove unnecessary "load" in order to facilitate learning.
## Types of Memory
Learning involves memory. For our purposes, human memory can be divided into two different
layers.
The first is called **long-term**.
It is where we store persistent information like our friends' names and our home address.
It is essentially unbounded
(barring injury or disease, we will die before it fills up)
but it is slow to access.
Our second layer of memory is called **short-term**. This is the
type of memory you use to actively think about things and is often called working memory. It is much faster, but also much
smaller: in 1956, George Miller estimated that the average adult's short-term memory could hold
**[7±2 items][wikipedia-7]** for a few seconds before things started to drop out.
This is why phone numbers are typically 7 or 8 digits long:
back when phones had dials instead of keypads,
that was the longest sequence of numbers most adults could remember accurately
for as long as it took the dial to go around and around.
More recent research suggests that short-term memory is actually even smaller than this.
Regardless of its exact size, which may differ across people and contexts, we know that
short-term memory is limited. This has important implications for teaching. If we present
our learners with large amounts of information, without giving them the opportunity to
practice using it (and thereby transfer it into long-term memory), they will not retain
the material as well as if we present small amounts of information interspersed with
practice opportunities. This is yet another reason why going slowly and using
frequent formative assessment is important.
::::::::::::::::::::::::::::::::::::::: instructor
For a semi-anonymous alternative to having learners write their score in the etherpad on the challenge below, you can create a list with numbers and have people add X's. This also makes it easier to see the score distribution.
*Example:*
```markdown
- 0
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12+
```
:::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::: challenge
## Test Your Working Memory
Try a short [test of working memory][memory-test] at [https://miku.github.io/activememory/][memory-test].
What was your score? If you are comfortable, share your answer
in the Etherpad.
If you are unable to use this activity, ask your Trainer to implement the analog version of this test.
This exercise should take about 5 minutes.
::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::: challenge
## Test Your Working Memory - Analog version (5 min)
Read the following list and try to memorise the items in it:
cat, apple, ball, tree, square, head, house, door, box, car, king,
hammer, milk, fish, book, tape, arrow, flower, key, shoe
Without looking at the list again, write down as many words from the list as
you can. How many did you remember? Write your answer in the Etherpad.
## Accessibility Adaptation
The list above may also be read aloud, slowly and clearly, by the Trainer hosting the session.
On your mark, ask learners to write down their answers where others cannot see them, then read the list
again so they can check their work.
::::::::::::::::::::::::::::::::::::::::::::::::::
Most people will have found they only remember 5-7 words. Those who
remember less may be experiencing distraction, fatigue, or (as we will learn shortly) "cognitive overload."
Those who remember more
are almost invariably deploying a *memory management strategy*.
## Strategies For Memory Management
Because short-term memory is limited, we can support learners by not flooding their
short term memory with too many separate pieces of information. Does this mean we should
teach fewer concepts? Yes! However, this is not the only tool in our toolbox.
We can also assist by providing strategies
and exercises to help them form the connections that will a) support them in holding more
things in short-term memory at once and b) begin to **consolidate** some concepts, moving them into
long-term memory.
### Chunking
Our minds can store larger numbers of facts in short-term memory by creating *chunks*, or
relationships among separate items, allowing them to be remembered as a single item.
For example, most of us will remember a word we read as a single item ("cat"), rather
than as a sequence of letters ("c-a-t").
Similarly, the pattern made by five spots on cards or die is remembered as a whole
rather than as five separate pieces of information.
::::::::::::::::::::::::::::::::::::::: challenge
## Improving Short-term Memory with Chunking
Repeat the memory exercise you did earlier, but this time,
try to form short stories or phrases, or a visual image, from the words you see.
How does this compare with your first attempt? Did your score increase, stay the same, or decrease? If you employed a strategy which one did you try?
This exercise should take about 5 minutes.
::::::::::::::::::::::::::::::::::::::::::::::::::
{alt='A list of words: cat, apple, ball, tree, square, head, house, door, box, car, king, hammer, milk, fish, book, tape, arrow, flower, key, shoe. Underneath, the same words are organised into boxes with related terms e.g. cat fish milk ball and apple flower tree"'}
Associating concepts
reduces the number of effective
items in your short-term memory, allowing you to keep
more information in your head at once.
You may have come across other mnemonic strategies, including some that rely on imagining a
"place" association for each item, e.g. a "memory palace." While slightly different from
chunking, this is another example of how
**connecting information can make it easier to remember**.
### Using Formative Assessment to Support Memory Consolidation
Formative assessment is a key component in helping learners solidify their understanding
and begin transferring ideas into long-term memory. Why? Because it engages the brain in
retrieving recently-learned information and actively applying it to solve a problem. This helps
to both reinforce and connect that new information in useful ways.
The limitations of short-term
memory are one reason why **assessments should be frequent**: short-term memory is limited not only in
space, but also in time. If you wait too long before deploying a formative assessment,
some of the information necessary to complete the task will already be forgotten. This time window can be very short,
especially if a lot of content is being taught at once! Be sure to remind learners about prior concepts
essential to a task. When you no longer need to remind them, this is a sign that your efforts in
supporting memory consolidation have worked!
### Group Work
**Elaboration**, or explaining your work, supports transfer
to long-term memory. This is one reason why **teaching is one of the most effective ways to learn**!
Group work can feel uncomfortable at first and consumes time in a workshop, but
learners often rate group work as a high point for both enjoyment and learning in a workshop.
This is also a great opportunity for helpers
to circulate and address lingering questions or engage with more advanced discussions.
### Opportunities for Reflection
Reflection is another tool that can help learners review things they have
learned, strengthen connections between them, and consolidate long-term memories. Like
formative assessment, **asking
learners for feedback can double as both a source of information and an effective consolidating
prompt**, as providing feedback demands some reflection on what has been learned. We will talk
more about methods for this in the next section. You may also wish to pause
and allow learners to write summary notes for themselves or otherwise ask them to review
what they have learned at various points in the workshop.
### Limit concepts
In the same vein as "going slowly," it is important to limit the number of
concepts introduced in a lesson. This can be hard! As you are reviewing a lesson
to teach, you will doubtless come across related concepts that are very useful, and
you may feel strongly motivated to sneak them in.
Planning your lesson with a concept map can help you
not only identify key concepts and relationships, but also to notice when you are trying to
teach too many things at once.
## Attention is a Limited Resource: Cognitive Load
Memory is not the only cognitive resource that is limited. Attention is constrained as well, which can
limit the information that enters short term memory in the first place as well as interfere with attempts
at consolidation.
While many people believe that they can "multi-task," the reality is that attention can
only focus on one thing at a time. Adding items that demand attention adds more things
to alternate between attending to, which can reduce efficiency and performance on all of them.
### The Theory of Cognitive Load
There are different [theories of cognitive load][wikipedia-cognitive-load]. In one of these, [Sweller](https://doi.org/10.1207/s15516709cog1202_4) posits that people have to attend to three types of things when they are learning:
- Things they have to think about in order to **perform a task** ("intrinsic").
- Mental effort required to **connect the task** to new and old information ("germane").
- **Distractions** and other mental effort not directly related to performing or learning from the task ("extraneous").
Cognitive load is not always a bad thing! There is plenty of evidence that *some* difficulty is desirable
and can increase learning. However, there are limits. Managing all forms of cognitive load, with particular
attention to extraneous load, can help prevent **cognitive overload** from impeding learning altogether.
One way to manage cognitive load as tasks become more complex is by using **guided practice**:
creating a structure that narrowly guides focus on specific skills and knowledge in a stepped fashion, with feedback at each step before
transferring attention to a new feature.
::::::::::::::::::::::::::::::::::::::::: callout
## Is Guided Practice "Hand Holding?"
An alternative to guided practice is a *minimal guidance* approach, where
learners are given raw materials (for example a text or reference) and
asked to explore and learn to solve problems on their own.
Minimal guidance is commonly found in many instructional strategies you may have encountered,
variously known as *constructivist*, *discovery*, *problem-based*, *experiential*
or *inquiry-based* learning.
These strategies are not without merit! Indeed, they can work exceptionally well with advanced learners.
However, they frequently fall flat, especially with novice audiences.
A landmark paper by [Kirshner et al.][kirschner-paper]
responds to the popularity and uneven success of minimal guidance, applying
cognitive load theory to understand why these strategies often fail.
Some people feel concerned that guided practice amounts to "hand-holding,"
implying that learners who receive support may never learn
to function independently. This view fails to account for the *additional cognitive
load* experienced by novices creating new connections while learning a task.
**Minimally-guided instruction requires learners to simultaneously
master a domain's factual content
AND its search and problem-solving strategies.** Fostering creativity and independence takes time.
Minimal guidance is intuitively appealing,
but that does not mean it always works.
::::::::::::::::::::::::::::::::::::::::::::::::::
::::::::::::::::::::::::::::::::::::::: challenge
## Mapping Cognitive Load
Look in the curriculum that you chose to prepare for this training and focus on one step or task
that learners will be asked to complete.
1. What concepts will learners need to understand and hold in short-term memory in order to complete this task?
2. Draw a concept map connecting these concepts. What relationships do learners need to understand to connect them?
3. How many of these concepts and relationships have been introduced since the previous step or exercise?
With a partner or in small groups, discuss what you have found. Are your learners at risk of cognitive overload at this point in your workshop? Why or why not?
This exercise should take about 15 minutes.
::::::::::::::::::::::::::::::::::::::::::::::::::
## Attention Management in Your Workshop
Carpentries lessons include small tasks arranged incrementally
which are intended to be completed together, through participatory live coding (a technique we will discuss in more detail
later in this training).
The choices you make as an Instructor may add to or subtract from your learners' cognitive load. Supporting memory consolidation
can reduce load later on in the workshop, as it reduces the effort of recalling forgotten material. You can also minimise
cognitive load by choosing formative assessments that are
narrowly focused and by considering potential distractions in what you display during instruction.
### Using Formative Assessments for Memory Management
There are
many different types of exercises that can focus attention narrowly and help to avoid cognitive overload.
Carefully targeted multiple choice questions can play this role.
A few more that you may wish to consider are:
- [Faded examples][wikipedia-faded]: worked examples with targeted details "faded" out -- essentially fill-in-the-blank programming blocks
- [Parson's Problems][wikipedia-parsons]: out-of-order code selection \& sorting challenges
- Labelling diagrams or flow charts (may also be organised as a fill-in-the-blank)
Beware of assessments that are too open-ended, as these are very likely to induce cognitive overload in novice learners! You may have experienced
some overload already when you were asked to create a concept map; this is why we do not recommend these as an activity for novice learners. Questions
that ask learners to both remember and synthesise or reason with new information are also risky. If you try out
a question and get a room filled with silence, you may need an icebreaker, you may need to re-teach... or you may only need a more narrowly focused question.
### What to Display
The Carpentries provides nicely formatted curricula for teaching. However, you may have noticed that you have not seen much, or perhaps
any of the Instructor Training curriculum during your time as a learner in this training. In most situations **we do not recommend
displaying Carpentries curriculum materials** to your learners while you teach.
The visual environment in a workshop should be focused on exactly what you are teaching and should mirror,
as closely as possible, exactly what you say.
This is because keeping track of distracting and contradictory sensory information adds to cognitive load.
The **[split-attention effect][wikipedia-split-attention]** describes the cognitive effort involved with
trying to assemble information from different modalities. Learning is most effective when visual
displays, text, and auditory information presented together
are the same, with minimal distractions.
For Carpentries workshops, this is why we ask Instructors to **speak commands as they type them on the screen**
while engaging learners in [participatory live coding](17-live).
One thing you may wish to consider adding to your (otherwise minimalist) visual environment, however, is a **running glossary of commands**
and other key terms. This can be maintained by a helper on a white board or an easel pad and will help learners readily access
items that may have already been dropped from short-term memory by the time they need them. In an online workshop, display of a glossary
is impractical because of severe limitations on screen space; however, a glossary can still be maintained in a collaborative document
for reference as needed.
## Summary
The process of learning is constrained by the limits of short-term memory. In order to move
new information into long-term memory, it must be actively applied, but activities that
make excessive demands on short-term memory are likely to induce cognitive overload and can
easily harm learner motivation. Instructional tools that expand short-term memory by
increasing connectivity (chunking) among new concepts can improve
outcomes for subsequent memory-intensive exercises. Formative assessments, when performed
frequently, help learners by prompting them to apply new content before it has been
overwritten. Faded examples or other types of guided practice
both minimise demands on short-term memory and offer context that
helps improve connectivity for future work, in which the "scaffolding" of contextual support
can be gradually removed.
Anything you can do to a) recognise and b) support learners in working with the
limitations of short-term memory will improve the effectiveness of your teaching.
[wikipedia-7]: https://en.wikipedia.org/wiki/The_Magical_Number_Seven,_Plus_or_Minus_Two
[memory-test]: https://miku.github.io/activememory/
[wikipedia-cognitive-load]: https://en.wikipedia.org/wiki/Cognitive_load
[kirschner-paper]: https://www.tandfonline.com/doi/pdf/10.1207/s15326985ep4102_1
[wikipedia-faded]: https://en.wikipedia.org/wiki/Worked-example_effect
[wikipedia-parsons]: https://en.wikipedia.org/wiki/Parsons_problems
[wikipedia-split-attention]: https://en.wikipedia.org/wiki/Split_attention_effect
:::::::::::::::::::::::::::::::::::::::: keypoints
- Most adults can store only a few items in short-term memory for a few seconds before they lose them again.
- Things seen together are remembered (or mis-remembered) in chunks.
- Cognitive load should be managed through guided practice to facilitate learning and prevent overload.
- Formative assessments can help to consolidate learning in long-term memory.
::::::::::::::::::::::::::::::::::::::::::::::::::
Generated concept map
The Skill created memory-learning.canvas with four topic groups and seventeen connections. This is the unchanged output exported by Obsidian, not a diagram redrawn by the evaluator. Open the image at full size to read the cards.

What the result shows
It chose the structure
The Skill chose four groups: memory and attention, cognitive load and guidance, building lasting connections, and designing the learning task. It condensed the lesson into sixteen concept cards, with separate title, exercise-summary and source-credit cards.
Important qualifications survived
The map keeps distinctions that matter. It does not call 7 +/- 2 a fixed memory limit, does not say all mental effort is harmful, and distinguishes the needs of novice and advanced learners.
One label is cut off
One arrow label is cut off at the left edge of the exported image. The card text itself is readable at full size, but this is a large map that needs zooming, not a compact one-screen summary.
How this example was scored
95/100 for this example only. Eight checks passed; readability was partly met because one arrow label is clipped. This is not a score for the whole collection or a guarantee for other notes. One real English lesson note, one json-canvas run and its native Obsidian output. The source already has headings; the Skill chose the map structure. The other four Obsidian Skills were not tested.
Five scoring dimensions
Passed checks receive 1 point, partly met checks 0.5, and failed checks 0. Checks outside this example are excluded, not counted as passes. The five dimension percentages are combined using the weights shown below.
Did it complete the task?
100.0%The map includes all nine topic areas checked against the lesson, from working memory and chunking to guided practice, display advice and a glossary. Seventeen labelled arrows link source-supported ideas, such as practice helping storage and formative assessment reinforcing consolidation. The Skill chose these links and the four groups; none were supplied as a target map.
Did it run without problems?
100.0%The canvas is valid JSON with unique identifiers, valid dimensions and resolved connection endpoints. It opens in Obsidian. The invocation completed, saved the requested editable file and left the supplied notes and Skill files unchanged. No command errors were recorded.
Did it preserve the supplied information?
75.0%The map keeps the important qualifications: 7 +/- 2 is a historical estimate, some difficulty can help learning, minimal guidance depends on learner experience, and the three load types belong to one theory. It does not present the lesson exercises as experiments performed in this test. All concept cards are readable at full size, but the label on the leftmost curved connector is partly cut off in the native export. The long diagram needs zooming on a small screen.
Did it do the documented task in this setup?
100.0%Retained execution shows the complete pinned json-canvas Skill and examples read before creation. The executor received the source note and request, not the previous result or scoring checklist. The unchanged canvas opens and its actual content is shown in the native Obsidian export. The edge clipping is recorded once under readability, not treated as a second application-compatibility failure.
Were the setup and use instructions clear?
100.0%The documented local canvas workflow was sufficient. The output credits The Carpentries, links to the original lesson and CC BY 4.0, and identifies the diagram as a summary and reorganisation.
Limits of this example
- Other four Obsidian Skills.
- Private or genuinely unstructured notes: the source is a real lesson with existing headings.
- Other AI applications/models, large vaults, mobile, Sync and installation from scratch.
- Independent verification of the lesson's scientific claims; this test checks fidelity to the supplied source.
Want to see more tests?
Tell us which module or task you would like us to test next.