⏳ This skill is pending AI review.

Scores will appear once the review pipeline completes.

version unknown

k12-check-for-understanding

anthropics@anthropics⭐ 628 stars· last commit 1mo ago· 2 open issues

>

Choose how to use this skill

You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.

1. Native installer

This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.

Do not guess an installer command or replace an existing version without reviewing the diff.

2. Complete package recommended

Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.

No complete ProSkills package is published for this listing yet.

3. Prompt-only

Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.

Need only the instruction file?

Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.

No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.

—/10

// RATINGS

⭐GitHub Stars
⭐⭐⭐⭐ 628 on GitHubGitHub ↗

Popular

🟢ProSkills Score
—
📍

Not yet listed on ClawHub or SkillsMP

// README

Agent Skills for K-12 Teachers

Source code and evaluation framework for the agent skills that are included with Claude for Teachers.

Included are four skills:

  • k12-lesson-plan-creation: Builds classroom-ready, standards-aligned lesson plans, optionally aligned to a teacher's curriculum
  • k12-lesson-differentiation: Adapts an existing lesson into tiered versions (below / at / above proficiency-level) and for specific student needs, keeping core content consistent across tiers
  • k12-lesson-prep: A prep partner for a lesson the teacher already has; works through the key student task with them and leaves a short teacher-only prep note
  • k12-check-for-understanding: Builds a 1–3 item formative check for a math standard, with distractors drawn from documented misconceptions and a teacher guide routing each response to a next step

Where noted, materials in this repo were co-developed between Anthropic and Learning Commons, who collaborated to help Claude create classroom materials that are grounded in academic standards and follow best practices from learning science research.

The skills are written to make effective use of the Learning Commons Knowledge Graph Claude connector that is included for all teachers using Claude for Teachers. This connector provides Claude with access to academic standards across all 50 states and progressions beneath them; adapt the skills if your environment does not have access to these tools.

Quick start

There are two options to use these skills:

In Claude for Teachers: If you have a Claude for Teachers account, there is no extra install needed. These skills are already installed for you.

In other places: Load the plugin or skills manually. For instance, in Claude Code:

git clone https://github.com/anthropics/k12-teacher-skills
claude plugin marketplace add ./k12-teacher-skills
claude plugin install k12-education@k12-teacher-skills

Layout

  • plugin/ - Main skill content bundled as a plugin; also includes teacher-focused 3rd party MCP servers that are available for users to enable
  • evals/ - Contains our evaluation framework and how to adapt them for your use case

// HOW IT'S BUILT

// FILES

9 files shipped with this skill. Inline text is the catalog copy. The upstream link is the source repository.

  • plugin/skills/k12-check-for-understanding/SKILL.md8.5 KB9d2eb4916899upstream ↗
    ---
    name: k12-check-for-understanding
    description: >
      Creates a Check for Understanding (CFU) for a math topic or standard: 1–3 targeted formative items
      whose distractors come from documented student misconceptions, plus a teacher guide routing each
      likely response to a specific next instructional step. Load BEFORE asking any clarifying question
      about grade, standard, or topic. This skill is for math; for other subjects, help as you normally
      would and don't mention the skill to the user. Triggers on asks to create, write, generate, or
      design a check for understanding, formative check, quick check, or exit ticket for a math concept
      or standard, and on implicit signals like "I need to see who got it before I move on." Not a
      quiz, test, or practice set — a CFU is read for what it reveals, not scored, so a numbered quiz
      or a bare answer key is a different job. Does not create a lesson (k12-lesson-plan-creation,
      which brings its own exit ticket), tier one (k12-lesson-differentiation), or prepare a teacher to
      teach one (k12-lesson-prep).
    license: "Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0. Complete terms in LICENSE and NOTICE."
    ---
    
    # K-12 Check for Understanding
    
    1–3 formative items for a standard the teacher is about to teach or just taught: scoped to one
    learning component within it, distractors drawn from documented student errors, plus a teacher guide
    saying what each response reveals and what to do. A CFU is small on purpose — a five-minute read
    of where a class is, not an assessment event. "The teacher" is the user you're talking with.
    
    **Asset-based language everywhere**, in both files and every chat message. No "weak," "fragile,"
    "can't," "behind," "deficient," "low," "struggling," "failing." Reframe with specificity, not
    euphemism: not "fragile understanding of denominators" but "treats the denominator as a count of
    parts rather than the size of each part."
    
    ---
    
    ## Keeping the teacher posted
    
    Keep the teacher posted in a sentence or two as you go, in a teacher's words: the topic, the grade,
    what students will be asked to do, what they're getting. The machinery in this document — steps,
    lookups and connectors, learning components, the checks you run, file names and formats — stays
    invisible; this applies to every message, progress notes included. "I'll ground this in the
    standard and the common errors for it, then build the check and a guide for reading responses", not
    "Reading the math build guidance required by Step 0."
    
    ---
    
    ## Step 0 — Route (silently)
    
    **Subject.** Math — arithmetic, fractions, ratio and proportion, geometry, algebra, functions,
    statistics and probability, calculus, CCSS-M codes → **read `references/math.md` NOW**. Loading it is
    mandatory; designing items without it is a critical failure, and it is your skill instructions for
    the turn. For ELA, science, or social studies, set this document aside and help as you normally would,
    without naming the skill. If the subject is ambiguous, ask in Step 1.
    
    **Connector.** Are Learning Commons KG tools (e.g. `find_standard_statement`) available? This decides
    Step 2; the skill works fully without them.
    
    ---
    
    ## Step 1 — Clarify
    
    At most one question, and only when the answer changes the items:
    
    - **Grade**, when not stated or inferable — "fractions" spans several grades, and the standard
      decides number values, representations, and what is on-grade. Ask before Step 2.
    - **Which standard**, when the topic maps to two plausible ones.
    
    Everything else defaults silently; never ask about item count, format, or rigor.
    
    ---
    
    ## Step 2 — Ground in the standard
    
    **Connected:** follow `references/learning-commons-kg.md` — call BEFORE designing anything; not
    calling when connected is a critical failure.
    
    **Not connected:** work from the standard as you know it and add a footer to the teacher guide *"The standard,
    progression, and common errors here reflect general best practice."* with nothing
    about what was or wasn't retrieved. Never invent KG citations or cite research you haven't seen.
    
    ---
    
    ## Step 3 — Select and confirm the learning-component focus
    
    A standard holds several separable ideas; scoping to one gives more actionable signal.
    
    1. Review **every** learning component returned in Step 2, if available.
    2. Select the **most assessable** — a genuine decision point in student understanding, probeable with
       1–3 items, separating students at different levels of understanding.
    3. **State the selection and rationale, then stop and ask.** What the component asks students to do
       in one plain sentence — a teacher's words, not the component's own wording — naming any constraint
       it carries; one line on why; one direct question. **Do not proceed to Step 4 until the teacher
       answers.**
    4. **If the teacher redirects** — a different component, or more than one — their choice becomes the
       scope anchor. Restate it and its rationale in one line and scope every item to it.
    
    Then say in a sentence or two what you're about to do (*"I'll pull the common errors research has
    documented for this standard, then build the check and a guide for reading the responses"*) — teacher
    language, no tool or file names.
    
    ---
    
    ## Step 4 — Build the check
    
    Follow the subject file's build section: coherence-map analysis, documented errors prioritized
    against the confirmed component, the misconception-to-stem map, item design by rigor, teacher-facing
    notes.
    
    **Data-use guardrail.** Curriculum guidance and misconception data inform item design and next-step
    wording internally only. **No curriculum, publisher, or research-source attribution appears anywhere
    in either file or in any chat message.** Write original items; never reproduce retrieved
    student-facing text, contexts, or teacher notes verbatim.
    
    ---
    
    ## Step 5 — Output
    
    In the turn Step 4 completes. Read `references/output.md` in full first. Then check whether your
    artifact tool can create an artifact from a published type; if it can, look at its type list. Take
    the first row that applies:
    
    | What you have | Do |
    |---|---|
    | **Slides** is in the type list | Read `rendering/claude-artifact.md` and make the check that way |
    | Slides isn't listed, or there is no such tool | Write both files to `$OUTPUT_DIR` |
    
    ---
    
    ## Step 6 — Verify
    
    **Starts only once both files exist.** Do not open `references/verification.md` during Step 4 or 5:
    it is a gate on the written artifact, not build guidance, and reading it early turns it back into a
    checklist you tick against what you meant to write.
    
    Now read it in full, read both files back from `$OUTPUT_DIR` (or the deck and the guide, on the
    slides path), and run its two gates on every item —
    whether more than one choice can be defended, and whether each distractor is a choice a real
    student's reasoning reaches. Fix what they find and rewrite the files. **Delivering a check that
    hasn't been through both gates is a critical failure.**
    
    ---
    
    ## Step 7 — Close
    
    Ask whether the check is what the teacher needs, offering at most two specific changes drawn from
    what this check is — a different rigor level, a second item on the same component, a
    constructed-response version. Not "let me know if you want changes." If they take one up, make it,
    rewrite both files, and ask again.
    
    **Plain language.** Never mention formats, file names, templates, how the documents were made, or the
    verification and what it changed: *"Here's the check and a guide for reading the responses"*, not
    *"I've written two HTML files."*
    
    ---
    
    ## Cross-cutting checklist
    
    - [ ] Focus confirmed with the teacher first, and every item scoped to it
    - [ ] The check has 1–3 items, no more
    - [ ] Labeled rigor matches the standard, format follows the rigor table, and numbers and
          representations respect the grade's range and the confirmed component's constraints
    - [ ] The coherence map grounds every prior-grade vs. on-grade route
    - [ ] One distractor is a prerequisite gap, one an on-grade confusion; no two rows route alike; every
          next step names a sub-skill, representation, or task type
    - [ ] Every student-facing sentence sits inside its grade band (K–2 ~8 words, 3–5 ~12 with at most
          one subordinate clause, 6–8 ~15), counted on the written stems and choices
    - [ ] Every figure carries `<title>`/`<desc>` with `role="img"` and `aria-labelledby` in both files;
          item text and choice order are identical between them
    - [ ] The student file has only prompts and choices/response space plus the name/date line; neither
          file carries a provenance marker or names a source
    
  • plugin/skills/k12-check-for-understanding/references/learning-commons-kg.md3.3 KB8ef60f5196a5upstream ↗
    
    # Learning Commons KG — call sequences for k12-check-for-understanding
    
    Used by Step 2 **only when the connector is available**; otherwise follow SKILL.md's not-connected
    branch. Calling the KG when connected is mandatory. Every call runs **before any item
    is designed**, and no raw results reach chat.
    
    ---
    
    ## Resolving the standard
    
    Infer the code from the topic, then call
    `find_standard_statement(code="<code>", academicSubject="Mathematics")`.
    
    - **A named state framework** (TEKS, SOL, OAS, CA, MA…): pass `jurisdiction="<state>"` and use that
      framework's codes rather than a CCSS proxy.
    - **Nothing returned:** retry with `keywords=[...]` and the distinctive content words.
    - **`subStandards` present:** the sub-part is usually the right anchor — a parent spanning several
      sub-parts is too broad for 1–3 items.
    
    Store the **verbatim statement** (meta block, never paraphrased), the **code**, and the
    **`caseIdentifierUUID`** every call below takes. Confirm it matches what the teacher asked for.
    
    ---
    
    ## Find learning components
    
    `find_learning_components_from_standard(caseIdentifierUUID)` returns the granular learning targets
    that make up the standard — the candidate scope anchors. Review **all** of them; the first returned
    is not the most assessable. Any constraints a component carries (number ranges, denominator sets,
    operation types, representational forms) bind item design. **If nothing is returned**, say so in the
    Step 3 proposal and select from the standard's own sub-parts instead, same rationale and same pause
    for confirmation.
    
    ---
    
    ## Find progression and coherence map
    
    `find_standards_progression_from_standard(caseIdentifierUUID, direction="backward")` and the same
    call with `direction="forward"` — both required. The results carry the **coherence map**, and the
    map, not the list of codes, grounds every prior-grade vs. on-grade route. Extract:
    
    1. **Prerequisite understanding** — what a student must already have. Keep the specific prior
       standard **codes**; the guide routes prerequisite-gap responses to a named one.
    2. **The unique demand of this standard** — what a student learns *at this grade* that they didn't
       know before and won't yet formalize later.
    3. **The on-grade vs. off-grade error line** — which errors mean a prerequisite gap versus an
       on-grade confusion. Decide it here, not at write-up time.
    
    ---
    
    ## Find misconceptions and guidance
    
    `find_misconceptions_for_standard(caseIdentifierUUID, subject="Mathematics")` returns
    research-grounded student errors plus instructional guidance. The **errors** drive distractor design;
    the **guidance** turns "revisit prior-grade content" into a named sub-skill or task type.
    
    Prioritize the 3–5 errors most relevant to the **confirmed learning component** rather than to the
    standard as a whole. If fewer than three are relevant, fill the remaining slots from your own
    knowledge; never manufacture a documented error. Record for each whether it is a **prerequisite gap**
    or an **on-grade confusion**.
    
    **Attribution:** none of these sources — publishers, curricula, research datasets alike — is named in
    either file or in any chat message (SKILL.md's data-use guardrail).
    
    ---
    
    **KG phase complete. Proceed to Step 3's focus proposal (or Step 4 if it is confirmed).**
    
    ---
    
    Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0
    
  • plugin/skills/k12-check-for-understanding/references/math.md6.7 KB937299887b9bupstream ↗
    # Math — k12-check-for-understanding
    
    Your full instruction set for the build (SKILL.md Step 4).
    
    ---
    
    ## 1. Aspect of rigor
    
    The standard's own verbs settle it — *understand*, *explain*, *represent* → **conceptual**;
    *fluently*, *compute*, *solve* → **procedural**; *use … to solve real-world problems* →
    **application**; spanning two → **Mixed**, saying which item covers which. Label it in the meta
    block; it drives item format, representation, and what a good distractor looks like.
    
    ---
    
    ## 2. Map documented errors to stem features
    
    Build this map from the errors prioritized in `references/learning-commons-kg.md` before writing any
    item. For each: *what would a student reasoning this way do here, and what must the stem contain to
    give them the chance?*
    
    | Error pattern | Prerequisite gap or on-grade? | Stem feature needed |
    |---|---|---|
    | [pattern] | [which] | [what must appear in the stem] |
    
    Internal working; it appears in neither file. **A distractor invented after the stem was
    written comes out merely wrong rather than plausible**, and discriminates nothing.
    
    ---
    
    ## 3. Item design
    
    Write 1–3 items; one that discriminates beats three that don't.
    
    | Rigor | Format |
    |---|---|
    | **Conceptual** | MC, multiple select, or CR — whichever reveals the understanding rather than the computation |
    | **Procedural** | CR is primary; at most one multiple-choice item |
    | **Application** | CR, with the student's approach and reasoning visible |
    
    - **Representation.** Conceptual K–8: at least one item includes or describes a diagram, number line,
      area model, or tape diagram — an all-symbolic conceptual item measures procedure instead.
      Procedural: at least one item shows work as a student would — a partially worked problem, a worked
      example with an error to find, or a show-your-work prompt. Any representational form the confirmed
      component names is the one to use.
    - **Number values** stay in the range the standard and grade expect: above it a conceptual check
      becomes an arithmetic obstacle course, below it a student can just count.
    - **Discrimination.** At least one item lets a student reach the correct answer by a faulty method,
      or exposes partial understanding in a "correct" response.
    - **Every item reveals thinking, not just correctness.** MC: every distractor maps to a named error
      from the map. CR: the prompt explicitly asks for the explanation, justification, or work — *"How do
      you know?"* at grade 3, *"Explain how you decided"* at grade 6.
    - **Figures state in words what they show** — "4 groups of 3 dots each", not "dice"
      (`references/output.md` has the markup). **And what a picture cannot say, the stem says:** *"Which
      expression shows the number of groups times the size of each group?"* is settled by figure plus
      stem; *"Which expression matches the picture?"* is not, because factor order is a convention the
      picture cannot carry.
    
    ### Multiple-choice items
    
    - Each distractor corresponds to one specific error from the map; prefer documented ones.
    - At least one reflects a **prerequisite gap** and at least one an **on-grade confusion**, so
      responses route to different next steps.
    - The key is not distinguishable by format, length, or specificity.
    - **A distractor is a place a real student's reasoning lands** — working this stem's numbers, the
      error arrives at that choice. Checked at `references/verification.md`, on the written files.
    
    ### Student-facing language sits at the grade's reading level
    
    Above-grade wording puts a second question inside the first, and the student lands in a distractor
    reporting a misconception they don't have. Applies to every string a student reads: stems, choices,
    directions, figure `<title>`/`<desc>`.
    
    - **Vocabulary.** Terms this grade's instruction teaches stay — *numerator*, *denominator*, *equal
      parts* belong in a grade 3 item. Every other word is the one the grade already has: *split into 4
      equal parts*, not *partitioned into fourths*; *find*, not *determine*. Standard wording is written
      for adults, so the meta block quotes it verbatim for the teacher and the item says the same idea in
      the students' words.
    - **Sentences.** One idea each, actor first, present tense, no idioms. **K–2** about 8 words; **3–5**
      about 12 and at most one subordinate clause; **6–8** about 15; **9–12** length comes from the
      mathematics. A sentence past its band carries something the mathematics didn't ask for — split it.
      A familiar context costs no reading; an unfamiliar one must be understood first.
    
    **Stems and choices are student-facing**: no annotation, symbol, label, bolding, or formatting that
    identifies the key, references an error pattern, or names a next step. Drafting them clean once makes
    the handout safe to hand over unedited.
    
    ---
    
    ## 4. Teacher-facing notes
    
    For each item:
    
    1. **What each response reveals** — one row per choice for MC, the expected response types for CR,
       describing the **observable behaviour**: what the student writes, draws, or circles, not an
       inferential label. "Adds the denominators as well as the numerators" is usable; "doesn't
       understand fraction addition" is not.
    2. **The next step.** Name which kind it is from the coherence map — **prerequisite gap** (prior-grade
       content not yet secure) or **on-grade confusion** (this standard's own demand) — then make it
       concrete from the retrieved guidance: the specific sub-skill, representation, or task type. A
       prerequisite-gap row routes to a **named prior standard code**, never "review earlier work."
    3. **What to look for across responses** — two or three sentences on whether a shared pattern points
       to one prior-grade gap several students can all be routed to, versus scattered on-grade confusions
       each needing different handling. **If two rows route alike, the data isn't working.**
    
    ---
    
    ## Meta-block content (templates: `references/output.md`)
    
    Six rows, in this order:
    
    - **`Standard`** — the verbatim statement retrieved in Step 2, never paraphrased, with its code.
    - **`Grade`** — the grade the items are written for.
    - **`Rigor`** — Conceptual / Procedural / Application / Mixed.
    - **`Learning goal`** — two sentences: what a student who has this can do, scoped to the confirmed
      focus, and how that shows up in these items.
    - **`Prerequisite standards`** — the 1–3 most relevant backward-progression standards, code plus a
      short gist, and the same codes the guide routes prerequisite-gap responses to.
    - **`Leads to`** — the 1–2 most relevant forward-progression standards, by code.
    - **Copyright:** never reproduce retrieved problem text, contexts, or teacher notes verbatim, and
      never name the curriculum, publisher, or research source.
    
    ---
    
    Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0
    
  • plugin/skills/k12-check-for-understanding/references/output.md12 KBccdc632d6dd5upstream ↗
    # Output — k12-check-for-understanding
    
    Read in full before writing (SKILL.md, Step 5).
    
    ---
    
    ## The two documents
    
    Every check produces **two separate self-contained `.html` files** — never one combined file, never
    markdown.
    
    - **`cfu_[topic-slug]_student.html`** — printable, distributable as-is: item prompts and answer
      choices (or response space for CR), a name/date line, nothing else. No meta block, no
      correct-answer marker, no source label, no next-step language.
    - **`cfu_[topic-slug]_teacher.html`** — meta block, focus and progression callouts, each item
      repeated in full, the correct answer, the interpretation guide, the what-to-look-for summary.
    
    `[topic-slug]` is a few lowercase hyphenated words: `equivalent-fractions`. **Item text is identical
    across the two files** — same prompts, choices, and order, figures included. Write it once and reuse
    it — that is what makes the handout safe to hand over unredacted.
    
    ---
    
    ## Writing the files
    
    Write both **into `$OUTPUT_DIR`** — `mkdir -p "$OUTPUT_DIR"`, then
    `"$OUTPUT_DIR"/cfu_equivalent-fractions_student.html` and the matching `_teacher.html`. That is where
    this runtime delivers from; a file written only to `/tmp` never reaches the teacher. Then list it and
    confirm both are present and non-trivial. With no `$OUTPUT_DIR`, present both
    documents inline in your reply, clearly separated. Both render standalone: styles inline in a
    `<style>` block, no external CSS, fonts, or scripts.
    
    ---
    
    ## Figures
    
    Read back as text an inline `<svg>` says nothing about what it shows, so it carries that in words —
    `<title>4 groups of 3 dots each</title>` and `<desc>Four squares in a row. Each square contains 3
    dots, for 12 dots in total.</desc>`, wired up as below.
    
    - **`<title>`** names the structure in the terms the item is read in — "4 groups of 3 dots each", not
      "dice"; **`<desc>`** adds what a student needs to work the item from that text alone. Both are
      student-facing, so both read at the grade's level.
    - **The description leaves the item's work undone.** Before writing it, name the one thing the item
      asks the student to do — count the parts, identify the coin, compare the two amounts — then write
      the text so a student working from it alone still has to do that thing. Where a line split into
      fourths asks which point is at 3/4, *"points A, B, C and D, one on each mark after 0"* leaves the
      counting to them; *"Point C is 3 jumps from 0"* has done it for them. Where the question is what a
      coin is worth, *"a small silver coin with a ridged edge"* leaves the recognition to them; *"a
      dime"* has not. What gives an item away is naming the property the question asks for — the count,
      the position, the identity, the comparison — which happens whether or not the answer itself
      appears in the text.
    - **It states only what the picture carries.** A comparison needs both things drawn: with a single
      coin on the page, *"one of the smaller coins in the jar"* reports something no reader can see.
    - `role="img"` with `aria-labelledby` on both makes a screen reader announce them, and identical
      markup goes in **both** files. Where the answer depends on a property only the figure
      carries, the stem names it too.
    
    ---
    
    ## No provenance or process talk
    
    Neither file says where anything came from: no `KG` or `Training` badge on a distractor row, no
    Source column, no badge in the meta block or on a callout, and no note that a lookup returned
    nothing. An empty progression result is a fact about the retrieval, not about the mathematics, so the
    guide states the prerequisite and next-step routing it does have and says nothing about the rest.
    Retrieved data still shapes every row; the file stops narrating it. (Third-party names stay out under
    SKILL.md's data-use guardrail.)
    
    Both files also read as though the check simply exists. `Confirmed with teacher`, `by teacher
    request`, `as you asked`, and any badge holding them are the conversation showing through the
    artifact — as is any trace of the checks you ran before delivering. Write the focus as a fact about
    the mathematics — *"Two components: identifying coin values, and adding a mixed set of coins"*.
    
    ---
    
    ## Styles
    
    Base block, in both files:
    
    ```css
    body{font-family:Georgia,serif;margin:40px auto;padding:0 24px;color:#1a1a1a}
    h1{font-size:1.35em;color:#2c2c2c;border-bottom:2px solid #444;padding-bottom:8px}
    .item-number{font-size:.85em;font-weight:bold;text-transform:uppercase;color:#555;margin-bottom:8px}
    .item-prompt{font-size:1.05em;line-height:1.7;margin-bottom:14px}
    .choices{list-style:none;padding:0;margin:0 0 16px}
    .choices li{padding:9px 13px;margin:7px 0;border:1px solid #767676;border-radius:4px}
    ```
    
    Student file adds:
    
    ```css
    body{max-width:700px}
    .name-line{margin:20px 0 32px}
    .name-line span{display:inline-block;border-bottom:1px solid #767676;min-width:220px;margin-left:6px}
    .item{margin:32px 0;page-break-inside:avoid}
    .response-line{border-bottom:1px solid #767676;height:32px;margin:8px 0}
    @media print{body{margin:0;padding:0 12px}}
    ```
    
    Teacher file adds:
    
    ```css
    body{max-width:800px;background:#fafaf8}
    h1{border-bottom-color:#d4a843}
    h2{font-size:.95em;text-transform:uppercase;color:#666;margin:20px 0 8px}
    .meta{background:#fff;border:1px solid #e0ddd5;border-radius:6px;padding:16px 20px;margin:20px 0;display:grid;grid-template-columns:200px 1fr;gap:0 12px}
    .meta-label,.meta-value{padding:8px 0;border-bottom:1px solid #f0ede5}
    .meta-label{font-weight:bold;color:#555}
    .meta-label:last-of-type,.meta-value:last-of-type{border-bottom:none}
    .item{background:#fff;border:1px solid #e0ddd5;border-radius:6px;padding:20px 24px;margin:24px 0}
    .correct-answer{color:#2d6a2d;font-weight:bold;margin-bottom:16px}
    table{width:100%;border-collapse:collapse;font-size:.92em}
    th{background:#f0ede6;text-align:left;padding:10px 12px;border-bottom:2px solid #ddd}
    td{padding:10px 12px;border-bottom:1px solid #eee;vertical-align:top;line-height:1.5}
    tr.correct-row td{background:#f0f7f0}
    .look-for,.lc-focus,.progression-ctx{padding:14px 18px;margin:20px 0;border-radius:6px;line-height:1.6}
    .look-for{background:#fef9ec;border-left:4px solid #d4a843}
    .lc-focus{background:#f1f8e9;border:1px solid #c5e1a5}
    .progression-ctx{background:#f0f4ff;border:1px solid #b8c8f0}
    .look-for strong,.lc-focus strong,.progression-ctx strong{display:block;margin-bottom:6px;font-size:.85em;text-transform:uppercase}
    .look-for strong{color:#7a5c00} .lc-focus strong{color:#2e7d32} .progression-ctx strong{color:#1a56c4}
    ```
    
    ---
    
    ## Student-facing template
    
    `<!DOCTYPE html>`, `<html lang="en">`, a head carrying `<meta charset="UTF-8">`, the viewport meta,
    `<title>CFU: [Topic]</title>`, and `<style>` with the base + student rules. Then:
    
    ```html
    <body>
    <main>
    
    <h1>[Topic/Concept]</h1>
    <div class="name-line">Name: <span>&nbsp;</span> &nbsp;&nbsp; Date: <span>&nbsp;</span></div>
    
    <!-- Repeat per item. Prompt and choices only — nothing else. -->
    <div class="item">
      <div class="item-number">Item 1</div>
      <div class="item-prompt">
        [Prompt text — identical to the teacher file. Where the answer turns on a property only the
        figure shows, this names that property.]
      </div>
      <svg viewBox="0 0 420 110" role="img" aria-labelledby="i1-t i1-d"
           xmlns="http://www.w3.org/2000/svg">
        <title id="i1-t">[what the figure shows, in the terms the item is read in]</title>
        <desc id="i1-d">[enough detail to work the item from this text alone]</desc>
      </svg>
      <ul class="choices">
        <li><strong>A.</strong> [Choice A]</li>
        <li><strong>B.</strong> [Choice B]</li>
        <!-- C, D likewise -->
      </ul>
      <!-- Constructed response instead: a few <div class="response-line"></div> in place of the <ul> -->
    </div>
    
    </main>
    </body>
    </html>
    ```
    
    ---
    
    ## Teacher-facing template
    
    **The prose fields are teacher notes, not assessment specs.** One thought per sentence, under about
    35 words; the plain phrase leads and the technical term follows in parentheses; the student is the
    subject. *"Students can locate a non-unit fraction on a number line by counting the total number of
    equal parts (denominator)…"* is a teacher's note; *"Reveal whether a student can coordinate both
    parts of locating…"* describes the design of the check.
    
    Same shell, with `<title>CFU: [Topic] — Teacher Guide</title>` and the base + teacher rules.
    
    ```html
    <body>
    <main>
    
    <h1>CFU: [Topic/Concept] — Teacher Guide</h1>
    
    <div class="meta">
      <!-- One label/value span pair per row, these six in this order. Contents per references/math.md. -->
      <span class="meta-label">Standard</span>
      <span class="meta-value">[CCSS code] — [verbatim standard language from the retrieval]</span>
      <span class="meta-label">Grade</span>
      <span class="meta-value">[Grade]</span>
      <span class="meta-label">Rigor</span>
      <span class="meta-value">[Conceptual / Procedural / Application / Mixed]</span>
      <span class="meta-label">Learning goal</span>
      <span class="meta-value">[Two sentences, one thought each, under about 30 words. First: what a student who has this can do — "Students can …", scoped to the confirmed focus. Second: how that shows up in these items. Not "students understand [standard]".]</span>
      <span class="meta-label">Prerequisite standards</span>
      <span class="meta-value">[1–3 codes, each with a short gist]</span>
      <span class="meta-label">Leads to</span>
      <span class="meta-value">[1–2 codes]</span>
    </div>
    
    <div class="lc-focus">
      <strong>Focus</strong>
      [Three or four short sentences, students as the subject, one thought each: what students do in a teacher's words; the constraints the confirmed focus carries (number range, denominator set, operation, representational form) in a sentence of their own; why this focus now. A fact about the mathematics, not how it was settled.]
    </div>
    
    <div class="progression-ctx">
      <strong>Progression &amp; coherence context</strong>
      [2–3 sentences: what prerequisite understanding this assumes, what makes this standard's demand distinctive at this grade, and how that decides whether a response routes prior-grade or on-grade.]
    </div>
    
    <!-- Repeat per item -->
    <div class="item">
      <!-- the same item-number / item-prompt / figure / choices markup as the student file, verbatim -->
      <h2>Correct answer &amp; interpretation guide</h2>
      <div class="correct-answer">✓ Correct answer: [Answer]</div>
    
      <table>
        <thead>
          <tr><th scope="col" style="width:22%">Response</th><th scope="col" style="width:40%">What it suggests</th><th scope="col">Suggested next step</th></tr>
        </thead>
        <tbody>
          <tr class="correct-row">
            <td>[Correct answer]</td>
            <td>Student [what they can now do, per the confirmed focus]</td>
            <td>[What extends or advances this]</td>
          </tr>
          <tr>
            <td>A. [Distractor]</td>
            <td>[What the student did, observably] — prerequisite gap: [prior-grade content not yet secure]</td>
            <td>[The specific prior standard code and the sub-skill to revisit]</td>
          </tr>
          <tr>
            <td>B. [Distractor]</td>
            <td>[What the student did, observably] — on-grade confusion: [which part of this demand]</td>
            <td>[Sub-skill or representation — distinct from every other row]</td>
          </tr>
          <!-- One row per remaining choice, same shape -->
        </tbody>
      </table>
    </div>
    
    <div class="look-for">
      <strong>What to look for across responses</strong>
      [2–3 sentences, grounded in the coherence map: whether a shared pattern points to one prior-grade gap those students can all be routed to, versus scattered on-grade confusions each needing different handling. Asset-based.]
    </div>
    
    </main>
    </body>
    </html>
    ```
    
    ---
    
    Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0
    
  • plugin/skills/k12-check-for-understanding/references/verification.md5.0 KBe7f1eec104a6upstream ↗
    # Verification — k12-check-for-understanding
    
    Read in full at Step 6, after both files are written and **before either is shown to the teacher**. Run both gates on every item; one that fails is fixed and the files rewritten before
    anything is delivered. The teacher never sees this — no narration, no "I checked and found…".
    
    **These gates run on files that already exist in `$OUTPUT_DIR`.** If you are reading this while still
    composing items, you are early: finish both files, then read them back and work from that text. By
    then you know which answer you meant, and a keyed item with two correct answers looks fine — so the
    gates read the item back **as text**, ask you to **write things down**, and run **after** the files
    exist.
    
    If a subagent is available, hand it only the student handout — no key, no guide, no context — and
    ask it to answer each item and report every choice it can defend; a fresh reader cannot rationalize
    toward the answer you meant.
    
    ---
    
    ## Gate A — exactly one defensible answer
    
    For each multiple-choice item, take every choice in turn — distractors included — and write one line
    stating **the strongest case that this choice is correct**, using only what the stem and its figure
    state. "Wrong because it isn't the key" is the gate failing silently. **Write the lines into a
    scratch file** (`/tmp/verify_notes.md`, not `$OUTPUT_DIR`): a case you have to type out is one you
    have to actually make. Then count the choices whose case survives.
    
    - **Exactly one** → the item passes.
    - **Two or more** → unanswerable as written, usually because the key rests on a property the stem
      never named. Repairs in order: (1) **name the property in the stem**, keeping the distractor — *"Which
      expression shows the number of groups times the size of each group?"* rules out the reversed factor
      order *"Which expression matches the picture?"* leaves open; (2) **remove the choice**.
    - **None** → either **a figure isn't described** (if you can't tell which choice is correct from the
      stem, the choices, and the figure's `<title>`/`<desc>`, neither can a student who can't see the
      picture — give it the missing text and rerun), or **the key is wrong** and the item needs
      reworking.
    
    Four recurring sources of a second defensible answer, worth checking directly:
    
    - **Factor order is a convention, not a fact** — `a × b` and `b × a` are the same number, so both are
      defensible unless the stem says which factor names the group count.
    - **Equivalent forms are one answer** — `1/2`, `0.5`, `2/4`, `50%` wearing four hats; two among the
      options means two correct answers.
    - **A right verdict reached by invalid reasoning** — where choices are whole statements (*"No,
      because…"*), several can carry the correct verdict and differ only in justification, which works
      only when the stem asks to be judged on the reasoning.
    - **Rounding, units, unstated precision** — a choice correct to the nearest whole unit is defensible
      against a key correct to the tenth unless the stem fixes the precision.
    
    ---
    
    ## Gate B — every distractor is a place a student's reasoning lands
    
    The reverse of Gate A: does the error each distractor is meant to reveal actually produce that
    choice? A distractor no described reasoning reaches is unpickable, and the student it was written for
    lands elsewhere with no row waiting.
    
    **First, compute the stem's actual quantities** — not what you intended, but the numbers as written:
    first differences, ratios column by column, totals, plotted coordinates. A few lines of `python3 -c`
    prints the property you were not looking for — a stem usually carries the one you chose it for and
    one you didn't.
    
    **Then, for each row of the interpretation guide**, apply the error it names to those quantities and
    state which choice it produces.
    
    - **That row's own choice** → passes.
    - **A different choice** → fails. Either the choice is wrong for that error or the row's claim is;
      fix whichever is mistaken, then check the row it lands on instead.
    - **The error can't be applied to this stem** → the distractor isn't reachable. Replace it with one
      this stem can produce.
    
    The worked failure: totals 45, 70, 95, 120 have first differences 25, 25, 25 — constant. A distractor
    reading *"the total paid does not increase by the same amount every time"* is false about that table,
    so a student applying the additive test reads constant differences and picks a different choice.
    Writing the differences down surfaces this; reading the distractor and nodding does not.
    
    Constructed-response items have no distractors, so Gate B applies to the response types the guide
    names: each is one a student could produce here.
    
    ---
    
    ## After the gates
    
    Fix what the gates found, rewrite both files, and re-run them on anything you changed — a
    repaired stem can break a distractor that was fine before. If an item still fails, **replace it
    rather than patch it again**. Both files are rewritten in place in `$OUTPUT_DIR`, so the fixed
    version is the first the teacher sees.
    
    ---
    
    Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0
    
  • plugin/skills/k12-check-for-understanding/LICENSE11 KB7a4a3ea2424cupstream ↗
    
                                     Apache License
                               Version 2.0, January 2004
                            http://www.apache.org/licenses/
    
       TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
    
       1. Definitions.
    
          "License" shall mean the terms and conditions for use, reproduction,
          and distribution as defined by Sections 1 through 9 of this document.
    
          "Licensor" shall mean the copyright owner or entity authorized by
          the copyright owner that is granting the License.
    
          "Legal Entity" shall mean the union of the acting entity and all
          other entities that control, are controlled by, or are under common
          control with that entity. For the purposes of this definition,
          "control" means (i) the power, direct or indirect, to cause the
          direction or management of such entity, whether by contract or
          otherwise, or (ii) ownership of fifty percent (50%) or more of the
          outstanding shares, or (iii) beneficial ownership of such entity.
    
          "You" (or "Your") shall mean an individual or Legal Entity
          exercising permissions granted by this License.
    
          "Source" form shall mean the preferred form for making modifications,
          including but not limited to software source code, documentation
          source, and configuration files.
    
          "Object" form shall mean any form resulting from mechanical
          transformation or translation of a Source form, including but
          not limited to compiled object code, generated documentation,
          and conversions to other media types.
    
          "Work" shall mean the work of authorship, whether in Source or
          Object form, made available under the License, as indicated by a
          copyright notice that is included in or attached to the work
          (an example is provided in the Appendix below).
    
          "Derivative Works" shall mean any work, whether in Source or Object
          form, that is based on (or derived from) the Work and for which the
          editorial revisions, annotations, elaborations, or other modifications
          represent, as a whole, an original work of authorship. For the purposes
          of this License, Derivative Works shall not include works that remain
          separable from, or merely link (or bind by name) to the interfaces of,
          the Work and Derivative Works thereof.
    
          "Contribution" shall mean any work of authorship, including
          the original version of the Work and any modifications or additions
          to that Work or Derivative Works thereof, that is intentionally
          submitted to Licensor for inclusion in the Work by the copyright owner
          or by an individual or Legal Entity authorized to submit on behalf of
          the copyright owner. For the purposes of this definition, "submitted"
          means any form of electronic, verbal, or written communication sent
          to the Licensor or its representatives, including but not limited to
          communication on electronic mailing lists, source code control systems,
          and issue tracking systems that are managed by, or on behalf of, the
          Licensor for the purpose of discussing and improving the Work, but
          excluding communication that is conspicuously marked or otherwise
          designated in writing by the copyright owner as "Not a Contribution."
    
          "Contributor" shall mean Licensor and any individual or Legal Entity
          on behalf of whom a Contribution has been received by Licensor and
          subsequently incorporated within the Work.
    
       2. Grant of Copyright License. Subject to the terms and conditions of
          this License, each Contributor hereby grants to You a perpetual,
          worldwide, non-exclusive, no-charge, royalty-free, irrevocable
          copyright license to reproduce, prepare Derivative Works of,
          publicly display, publicly perform, sublicense, and distribute the
          Work and such Derivative Works in Source or Object form.
    
       3. Grant of Patent License. Subject to the terms and conditions of
          this License, each Contributor hereby grants to You a perpetual,
          worldwide, non-exclusive, no-charge, royalty-free, irrevocable
          (except as stated in this section) patent license to make, have made,
          use, offer to sell, sell, import, and otherwise transfer the Work,
          where such license applies only to those patent claims licensable
          by such Contributor that are necessarily infringed by their
          Contribution(s) alone or by combination of their Contribution(s)
          with the Work to which such Contribution(s) was submitted. If You
          institute patent litigation against any entity (including a
          cross-claim or counterclaim in a lawsuit) alleging that the Work
          or a Contribution incorporated within the Work constitutes direct
          or contributory patent infringement, then any patent licenses
          granted to You under this License for that Work shall terminate
          as of the date such litigation is filed.
    
       4. Redistribution. You may reproduce and distribute copies of the
          Work or Derivative Works thereof in any medium, with or without
          modifications, and in Source or Object form, provided that You
          meet the following conditions:
    
          (a) You must give any other recipients of the Work or
              Derivative Works a copy of this License; and
    
          (b) You must cause any modified files to carry prominent notices
              stating that You changed the files; and
    
          (c) You must retain, in the Source form of any Derivative Works
              that You distribute, all copyright, patent, trademark, and
              attribution notices from the Source form of the Work,
              excluding those notices that do not pertain to any part of
              the Derivative Works; and
    
          (d) If the Work includes a "NOTICE" text file as part of its
              distribution, then any Derivative Works that You distribute must
              include a readable copy of the attribution notices contained
              within such NOTICE file, excluding those notices that do not
              pertain to any part of the Derivative Works, in at least one
              of the following places: within a NOTICE text file distributed
              as part of the Derivative Works; within the Source form or
              documentation, if provided along with the Derivative Works; or,
              within a display generated by the Derivative Works, if and
              wherever such third-party notices normally appear. The contents
              of the NOTICE file are for informational purposes only and
              do not modify the License. You may add Your own attribution
              notices within Derivative Works that You distribute, alongside
              or as an addendum to the NOTICE text from the Work, provided
              that such additional attribution notices cannot be construed
              as modifying the License.
    
          You may add Your own copyright statement to Your modifications and
          may provide additional or different license terms and conditions
          for use, reproduction, or distribution of Your modifications, or
          for any such Derivative Works as a whole, provided Your use,
          reproduction, and distribution of the Work otherwise complies with
          the conditions stated in this License.
    
       5. Submission of Contributions. Unless You explicitly state otherwise,
          any Contribution intentionally submitted for inclusion in the Work
          by You to the Licensor shall be under the terms and conditions of
          this License, without any additional terms or conditions.
          Notwithstanding the above, nothing herein shall supersede or modify
          the terms of any separate license agreement you may have executed
          with Licensor regarding such Contributions.
    
       6. Trademarks. This License does not grant permission to use the trade
          names, trademarks, service marks, or product names of the Licensor,
          except as required for reasonable and customary use in describing the
          origin of the Work and reproducing the content of the NOTICE file.
    
       7. Disclaimer of Warranty. Unless required by applicable law or
          agreed to in writing, Licensor provides the Work (and each
          Contributor provides its Contributions) on an "AS IS" BASIS,
          WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
          implied, including, without limitation, any warranties or conditions
          of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
          PARTICULAR PURPOSE. You are solely responsible for determining the
          appropriateness of using or redistributing the Work and assume any
          risks associated with Your exercise of permissions under this License.
    
       8. Limitation of Liability. In no event and under no legal theory,
          whether in tort (including negligence), contract, or otherwise,
          unless required by applicable law (such as deliberate and grossly
          negligent acts) or agreed to in writing, shall any Contributor be
          liable to You for damages, including any direct, indirect, special,
          incidental, or consequential damages of any character arising as a
          result of this License or out of the use or inability to use the
          Work (including but not limited to damages for loss of goodwill,
          work stoppage, computer failure or malfunction, or any and all
          other commercial damages or losses), even if such Contributor
          has been advised of the possibility of such damages.
    
       9. Accepting Warranty or Additional Liability. While redistributing
          the Work or Derivative Works thereof, You may choose to offer,
          and charge a fee for, acceptance of support, warranty, indemnity,
          or other liability obligations and/or rights consistent with this
          License. However, in accepting such obligations, You may act only
          on Your own behalf and on Your sole responsibility, not on behalf
          of any other Contributor, and only if You agree to indemnify,
          defend, and hold each Contributor harmless for any liability
          incurred by, or claims asserted against, such Contributor by reason
          of your accepting any such warranty or additional liability.
    
       END OF TERMS AND CONDITIONS
    
       APPENDIX: How to apply the Apache License to your work.
    
          To apply the Apache License to your work, attach the following
          boilerplate notice, with the fields enclosed by brackets "[]"
          replaced with your own identifying information. (Don't include
          the brackets!)  The text should be enclosed in the appropriate
          comment syntax for the file format. We also recommend that a
          file or class name and description of purpose be included on the
          same "printed page" as the copyright notice for easier
          identification within third-party archives.
    
       Copyright [yyyy] [name of copyright owner]
    
       Licensed under the Apache License, Version 2.0 (the "License");
       you may not use this file except in compliance with the License.
       You may obtain a copy of the License at
    
           http://www.apache.org/licenses/LICENSE-2.0
    
       Unless required by applicable law or agreed to in writing, software
       distributed under the License is distributed on an "AS IS" BASIS,
       WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
       See the License for the specific language governing permissions and
       limitations under the License.
  • plugin/skills/k12-check-for-understanding/NOTICE320 B4b912282e26bupstream ↗
    Copyright 2026 Anthropic, PBC
    Copyright 2026 Learning Commons
    
    Acknowledgment
    We wish to acknowledge that this skill was refined through evaluation and feedback from Student Achievement Partners (SAP). SAP's involvement does not imply endorsement or guarantee of how the skill is used, modified, or built on downstream.
    
  • plugin/skills/k12-check-for-understanding/rendering/claude-artifact.md1.7 KBf47f9953e6adupstream ↗
    # Make the check as Claude slides
    
    Use this when Step 5 found a **Slides** type. The check becomes a slide deck the teacher projects,
    and the teacher guide becomes a Claude Doc.
    
    `references/output.md` still sets the content: the items and their exact wording and order, what a
    figure's description may say, the guide's sections, and no provenance. Its file names, styles and
    HTML templates are only for the file path.
    
    ## The deck
    
    Create it from the Slides type with the topic as its title, then follow the Slides type's own
    instructions for writing and formatting it. Give each item its own slide: the prompt, any figure,
    and the choices, large enough to read from the back of the room. Nothing else goes on the deck — no
    answers, no notes for the teacher.
    
    Draw a figure that has to be exact with a short Python script (matplotlib, or plain SVG), and give it
    alt text by the same rule as a figure's `<desc>`: enough to work the item, never the item's work done.
    
    ## The teacher guide
    
    Make it a Claude Doc, following Claude Docs' own instructions, with the sections of the teacher
    template: each item word for word as on its slide, its answer and interpretation table, and what to
    look for. Without Claude Docs functionality, write the guide as the teacher `.html` file instead.
    
    ## If it can't be made
    
    If the deck can't be created or written, write both `.html` files as `references/output.md` says and
    don't mention slides.
    
    ## Hand it over
    
    Give the teacher the links in plain words. Don't talk about tools or formats. To revise, change the
    affected slide and its place in the guide, and nothing else.
    
    ---
    
    Copyright 2026 Anthropic, PBC · Copyright 2026 Learning Commons · SPDX-License-Identifier: Apache-2.0
    
  • README.mdupstream ↗

    Contents are not inlined in the catalog. Open the upstream file.

Plain-text copy, no login: /skill-bundles/anthropics-k12-teacher-skills-plugin-skills-k12-check-for-understanding.md

// REPO STATS

628 stars
2 open issues
Last commit: 1mo ago