mylab-pearson
Skill1 starsChanged 2 months agoMITView on GitHub
Work a Pearson MyLab / MyLab Statistics / MathXL homework set or quiz — launch it through the popup, read each question, compute, fill the answer boxes, and submit. Use whenever Max points at a MyLab, MyLab Stats, Mastering, MathXL or Pearson assignment, says a stats homework is "in MyLab", asks to do/finish/check a numbered set like "2.3" or a chapter review quiz, or when an assignment turns out to be hosted at mylabmastering.pearson.com / mylab.pearson.com / tdx.acs.pearson.com. Covers the popup launch, promoting the player out of its iframe, the custom math answer boxes, and reading graphs from their alt text. Not for MyOpenMath (myopenmath.com) — that is the `myopenmath` skill.
In this repository4 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- canvas-homeworkSkill
- canvas-quizSkill
- myopenmathSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
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 10
tooling/MYLAB_PEARSON.md: Read it before the first interaction. - Very specific: line 35
Promote the player to top-level: before clicking *Get started*: location.href = …
- Distinctive: line 75
Keep the figure data and your previous answers to hand.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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
MyLab / Pearson homework
Drives a Pearson MyLab (MathXL) assignment end to end with the Claude in Chrome
tools. The full runbook — every trap, with the fix — is
tooling/MYLAB_PEARSON.md. Read it before the first interaction; what follows
is the shape of the job.
The page helper is tooling/mylab-pearson.js.
Which skill
myopenmath.com→ themyopenmathskill. Different machinery (MathQuill, hash routing, per-question submits).mylabmastering.pearson.com/mylab.pearson.com→ this skill.
Ask first
Reading and computing is ordinary help. Opening an attempt, typing into boxes, and clicking Check answer each need Max's go-ahead. One ask covers a homework set with unlimited attempts. A timed or limited-attempt quiz is different: the irreversible step is opening it, so quote the attempt count, time limit and due date and ask before you start. Never enter credentials.
Steps
- Get the tab into the MCP group. Clicking the assignment fires a popup that lands outside the group — ask Max to drag it in. Never let the group reach zero tabs, it dissolves (§1).
- Promote the player to top-level before clicking Get started:
location.href = document.getElementById('ctl00_ctl00_InsideForm_MasterContent_PlayerHtml5').srcYou land ontdx.acs.pearson.comwith a real DOM instead of being screenshot-only. This is the move that makes the whole run fast (§2). - Paste
tooling/mylab-pearson.js, then__ML.hookPopups(). - Scope it: question count, points, due date, and whether tries are limited.
- Per part:
__ML.q()to read →__ML.openFig()+__ML.fig()for any figure, or__ML.table()+__ML.stats()for a data table (§14) → compute → click the left edge of the box and type → verify with__ML.fields()→ wait 1s (§16) →__ML.hit(/^Check answer/). Batch the whole part (fill, wait, check, screenshot) into onebrowser_batchcall rather than one tool call per step. - Confirm with
__ML.score(), then__ML.hit(/^Next/). If the question finished with partial credit, the default is to click "See similar" and redo it on a fresh instance — but MyLab keeps whichever attempt's score is most recent, not the higher one (§18), so skip the retry on a part you can't verify with confidence before submitting — histogram parts now can be, via__ML.histAlt()/__ML.histBars()(§20) — or on a question already scoring 0.8+. - Write the study note into
output/study-notes/<course>-<set>.md, same format as themyopenmathskill's step 6: atopics:line, then Formulas, Strategies, Traps, Question index. Write it right after the last question, while the work is still in context.
The five that actually cost something
- Read graphs from the alt text, not the SVG.
__ML.fig()returns a screen-reader description listing every plotted coordinate exactly. Parsing the layered SVGs is a last resort. - Click the left ~22% of an answer box. The right side is a math-palette button; hit it and your keystrokes vanish into a focused button while the box stays empty.
- Clear a box with
cmd+athenDelete. Backspace corrupts it silently (10→ typing9produced90). - Parts are revealed one at a time and later parts depend on earlier ones. Keep the figure data and your previous answers to hand; re-query field coordinates and scroll before clicking, because new parts appear below the fold.
- Tries are per part (3 for free response, often 2 for multiple choice), and a wrong answer disables Check answer until you change something. "Unlimited per question" on the overview does not mean unlimited per part.
Stats content traps
- Class limits are
(lower, lower+width−1), not(lower, lower+width)— midpoint 5 with width 10 means the class 0–9. - A frequency polygon's first and last points are zero anchors, not classes.
- On an ogive, "p% below" reads x at y = p/100; "p% above" reads x at y = 1−p/100.
- Ties are real — when two candidates tie, MyLab wanted the later one.
- Read each box's rounding instruction literally. "Round to two decimal
places" wants
__ML.preciseRound(); "Do not round" wants the exact__ML.stats()value — don't blanket-round every box in a question the same way (§15). - A rejected mean/median that survives triple-checked extraction may be a
data-generation quirk, not your error: the displayed table can hide decimal
precision. A
+$0.01retry has worked twice for a mean; for a median, don't guess blind — use "See similar" instead (§17, §18 in the runbook). - "Help me solve this" can regenerate the question's data from a single stray click on the panel — even without clicking "Continue" inside it — silently invalidating already-correct parts. Avoid it; re-extract fresh and re-verify prior parts if it's triggered by accident (§19).
- Histogram multiple-choice options carry a hidden exact
aria-labelon adivancestor of each thumbnail —__ML.histAlt()returns them in the same order as the visual A/B/C/D layout, and__ML.histBars(alt)parses one into[x, relFreq]pairs to diff against your own computed bins. This is exact, not a guess — use it instead of eyeballing peak position/tail shape, which was wrong about as often as right (§20). - Verify a data table's shape before trusting
__ML.table()— it assumes a duplicate footer row and single-column-per-variable layout that not every table has. Checkrows.lengthagainst the question's stated sample size, and check headercolSpanfor side-by-side repeated column groups, before computing anything (§21).
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.

