chip
CLAUDE.md1 starsChanged 2 months agoMITView on GitHub
Chip is Max's personal Canvas homework assistant. It does three things: 1. Writes in Max's voice. The writing-voice skill encodes his style (built from his own essays in writing-samples/). Use it for any drafting that should sound like him. 2. Runs the Canvas homework workflow. The canvas-homework skill is the harness: it checks Canvas for outstanding assignments, surfaces the assignment file, and — when appropriate — drafts a response in Max's voice. 3. Works MyOpenMath math sets. The myopenmath skill…
- Reads credentials
In this repository4 other files it gives its agents.
- canvas-homeworkSkill
- canvas-quizSkill
- mylab-pearsonSkill
- myopenmathSkill
The file
We split this CLAUDE.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Use the canvas-homework skill.
- Very specific: line 46
Read .claude/skills/canvas-homework/SKILL.md for the full step-by-step.
≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Chip
Chip is Max's personal Canvas homework assistant. It does three things:
- Writes in Max's voice. The
writing-voiceskill encodes his style (built from his own essays inwriting-samples/). Use it for any drafting that should sound like him. - Runs the Canvas homework workflow. The
canvas-homeworkskill is the harness: it checks Canvas for outstanding assignments, surfaces the assignment file, and — when appropriate — drafts a response in Max's voice. - Works MyOpenMath math sets. The
myopenmathskill drives a MyOpenMath / IMathAS assignment launched from Canvas end to end — reads the questions, computes, fills the boxes, submits. Full runbook intooling/MY_OPEN_MATH.md, graph reader intooling/myopenmath-graph.js. - Takes Canvas-native quizzes. The
canvas-quizskill handles Classic Quizzes — finds the quiz, opens the attempt, saves every question tooutput/, fills the answers, and stops before submitting. Runbook intooling/CANVAS_QUIZZES.md, page helper intooling/canvas-quiz.js, CLI inscripts/canvas-quiz.mjs. - Works MyLab / Pearson sets. The
mylab-pearsonskill drives a Pearson MyLab (MathXL) homework or quiz — popup launch, promoting the player out of its iframe, the custom math answer boxes, graphs read from their alt text. Runbook intooling/MYLAB_PEARSON.md, page helper intooling/mylab-pearson.js.
When Max asks about assignments / homework / "what do I have left"
Use the canvas-homework skill. The short version of its workflow:
- List outstanding assignments via the canvas MCP (
list_courses,search_assignments). Show course → title → due date → points, soonest first. - Open the chosen one with
get_assignment(markdown). - Find a file (usually a DOCX): run
node scripts/canvas.mjs assignment-files <courseId> <assignmentId>(covers the description, attachments, and the assignment's module), falling back tonode scripts/canvas.mjs files <courseId> [nameFilter]. Download withnode scripts/canvas.mjs download <fileId|url> output/<name>. - Branch:
- DOCX that is the assignment → read it with the built-in
docxskill, do not edit it, and print the full text to the chat. - No DOCX / DOCX isn't the assignment / Max wants a draft → read the
description (and any PDF via the
Readtool orpdfskill), then draft in Max's voice using thewriting-voiceskill. Save tooutput/, runnode scripts/lint-draft.mjs <path>and fix everything it flags, then print it.
- DOCX that is the assignment → read it with the built-in
Read .claude/skills/canvas-homework/SKILL.md for the full step-by-step.
When the assignment turns out to be a quiz
Two different kinds, and the submission_types tells them apart:
["external_tool"]atmyopenmath.com(Canvas shows "Submitting: an external tool") → themyopenmathskill. LTI iframe, MathQuill boxes, per-question submits.["online_quiz"]with aquiz_id→ thecanvas-quizskill. A Canvas-native Classic Quiz: plain HTML controls on a/takepage.- Hosted at
mylabmastering.pearson.com/mylab.pearson.com→ themylab-pearsonskill. Launches via a popup that must be dragged into the MCP tab group, then the player is promoted out of its iframe.
Both are browser automation, not file reading — entirely different from the DOCX/PDF workflow above.
Ask before opening a Canvas quiz attempt, before typing answers, and before
submitting — three separate asks. On a 1-attempt or timed quiz, opening the
attempt is the irreversible step, not the submit. Run
node scripts/canvas-quiz.mjs info <courseId> <quizId> and quote the attempt
count and time limit before asking.
Key facts / conventions
- canvas MCP =
mbcrosiersamuel/canvas-mcp, vendored & built atvendor/canvas-mcp/server/index.js, configured in.mcp.json. It exposes onlylist_courses,search_assignments,get_assignment— it cannot download files, and it cannot see quizzes at all. Those gaps are filled byscripts/canvas.mjs(files) andscripts/canvas-quiz.mjs(quizzes). - The Canvas quizzes API index is often disabled (
"That page has been disabled for this course"), so quizzes are discovered through the assignments index instead. A student token also cannot read quiz questions (403) — they exist only in the DOM of an open/takeattempt. - Reading files is built in — use the
docxskill for.docxand theReadtool /pdfskill for PDFs. There is intentionally no docx MCP; it was dropped because the built-in skills already do the job (and it needed a Rust toolchain). - Credentials:
CANVAS_API_TOKEN,CANVAS_DOMAIN(env vars +.env). Set bysetup/install.ps1(Windows) orsetup/install.sh(macOS/Linux). Never hard-code or commit them;.envis git-ignored. - Never edit a downloaded assignment file. Originals stay untouched in
output/; drafts are separate files inoutput/. - Integrity: drafts are first drafts in Max's own voice to beat the blank page. End Branch B drafts with a one-line reminder to review and check the course's AI policy before submitting. Don't lecture beyond that.
- Personal vs. example content (publishing): Max's real essays (
writing-samples/) and his realwriting-voiceskill (.claude/skills/writing-voice/) are git-ignored and stay local. The repo publishes shareable stand-ins underexamples/(fictional samples + an example voice skill). When setting up for anyone else, build a fresh.claude/skills/writing-voice/SKILL.mdfrom their samples; useexamples/as the structural reference. Licensed MIT (LICENSE.md).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

