Explain

Skill IvanGomezMartinez/MatStudyLab/.agents/skills/explain

Deep-study a MATLAB script and write explain_<stem>.md without changing .m files.From its SKILL.md

Install

npx -y skills add IvanGomezMartinez/MatStudyLab --skill explain

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.2 KB, 766 tokens by cl100k_base, as published. Nobody here has run it

Produce explain_<stem>.md in explain/ for one script per session. Read-only on .m files — never edit codes/, new/, modify/, or import/ in place.

Read LORE.md, AGENTS.md, docs/spec.md, and CONTEXT.md before acting. Use docs/templates/explain-doc.md for structure.

Step 0 — Bootstrap (mandatory)

Read and execute .agents/skills/matstudylab-bootstrap/SKILL.md before any other step. Do not proceed until bootstrap completes (sync, skip, or failed-with-continue).

Completion criterion: bootstrap outcome identified (fresh, synced, or failed-with-continue).

Step 1 — Parse invocation

User input	Behavior
`/explain`	Resolve a single `.m` under `explain/`, or list choices if several
`/explain <script.m>`	Study that script (paths under `explain/`, `codes/`, `new/`, `modify/`)
`/explain <script.m> <question>`	Same + answer the question in Your questions

No grill-me — didactic, one-way mode.

Completion criterion: target script path is identified.

Helper:

./scripts/explain-resolve.sh list
./scripts/explain-resolve.sh resolve <hint>

Step 2 — Read context

Read LORE.md for companion language and lab context. Read the target .m and its base companion .md when present (codes/ is read-only context).

Apply zoom-out when the script is long, has non-obvious dependencies or toolboxes, or the user gave a hint — read related files under the same bundle before writing.

Completion criterion: script purpose and optical context are understood.

Step 3 — Write explain doc (explain/ only)

Resolve output path:

./scripts/explain-resolve.sh resolve <hint>

Scaffold if needed, then fill content in the user's language:

./scripts/explain-resolve.sh scaffold <hint> "<optional question>"

Required sections (see docs/templates/explain-doc.md):

Possible improvements (not implemented) — ideas only, no code changes
MATLAB concepts — plain-language teaching
Your questions — when the user provided a hint

Completion criterion: explain_<stem>.md exists under explain/ with all required sections; no .m file was modified.

Step 4 — Recommend `/accept explain`

If a matching bundle exists in codes/<type>/<bundle>/, recommend:

/accept explain

or /accept explain <type> <bundle> when scope should be one bundle. This copies explain_<stem>.md into the catalog (see .agents/skills/accept/SKILL.md).

If the script has no catalog bundle yet, suggest /build or /new first, then /accept for the staging bundle.

Completion criterion: user knows the next command to attach the study doc.

Safety (non-negotiable)

Write only under explain/.
Never modify .m files.
Never edit codes/ in situ.

Reference

Wayfinder spec: .scratch/matstudylab/issues/09-grilling-comando-explain.md
Explain template: docs/templates/explain-doc.md
Handoff: use .agents/skills/handoff/SKILL.md to compact context before generating the doc when zoom-out applies

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.