@graffiticode/l0175 0.2.0 → 0.3.1

package/spec/rag-examples-design.md

@@ -0,0 +1,74 @@
+<!-- SPDX-License-Identifier: CC-BY-4.0 -->
+# L0175 RAG anchor set — design
+
+_Status: DESIGN (authoring of the golden pairs is a follow-up). Revised 2026-06-30._
+
+## Why this exists
+
+The reported failure: asked for **c1-t9, Task Model 3 (two-part EBSR)**, the generator drifts to
+TM4 (hot-text) / TM5 (short-text). The instruction-side fixes (the served collision warning, the
+generated per-target task-model table, and the compiler's `task-model` guard) make the rule
+explicit and enforceable. RAG adds the third leg: a **positive worked example** of the exact shape
+the request wants, so retrieval biases generation toward it instead of toward the more salient
+T4/T11 numbering.
+
+The drift is between **siblings within one target** (t9 EBSR vs hot-text vs short-text). An abstract
+per-target renumbering rule is exactly what an LLM under-weights; a retrieved exemplar of
+"c1-t9, TM3, two-part EBSR" is the strongest available anchor. The embedding infra already supports
+this — `src/embedding.ts` emits `target:c1-t9`, `item:ebsr`, `shape:two-part`, and now
+`task-model:3`; `extractQueryFacets` pulls target + item type + an explicit/derived task-model
+number from the query. The corpus just lacks positive output-side exemplars.
+
+## What the set should be
+
+1. **Paired golden examples (prompt → canonical L0175 program), not prompt-only.** `examples.md` is
+   prompt-only; disambiguating siblings needs the *output* shape. Store paired examples the console
+   RAG can index via `buildEmbeddingArtifacts({ prompt, code })` — e.g. a `spec/examples/` directory
+   of `<id>.gc` programs each with a sibling `<id>.prompt.md`, or an L0175 `training_examples.json`
+   array of `{ prompt, code }`.
+
+2. **Contrastive siblings — same passage, vary only the task model — for every target** (parity, not
+   t9-only). Holding the passage + dimension fixed and changing only the task model makes the
+   vectors/facets separate cleanly on item type / shape, so retrieval discriminates instead of
+   blurring:
+
+   | Target | Task models to anchor (item type) |
+   |--------|-----------------------------------|
+   | `c1-t9` (validate first — the observed failure) | tm3 ebsr · tm4 hot-text · tm5 short-text · tm1 multiple-choice · tm2 multi-select |
+   | `c1-t1`, `c1-t8`, `c1-t10` | tm1 multiple-choice · tm2 multi-select · tm3 hot-text |
+   | `c1-t4`, `c1-t11` | tm1 ebsr · tm2 hot-text · tm3 short-text |
+
+   Every target × task-model cell gets ≥1 anchor. No target ships without anchors. (Source of truth
+   for which cells exist: `spec/targets.json` → `targets.<id>.taskModels`.)
+
+3. **Tagged with the design signature `embedding.ts` emits** — `target:<t>`, `item:<type>`,
+   `shape:two-part|single-part`, `dimension:<d>`, `standard:<s>`, `dok:<d>`, and **`task-model:<n>`**
+   (derived from target + item type via `targets.json`). The console filters/boosts on these facets;
+   `target:c1-t9` + `task-model:3` together select the EBSR exemplar over its siblings.
+
+4. **Each prompt phrased both ways** — by number ("task model 3", "TM3") **and** by item type
+   ("two-part EBSR") — so the query embedding and `extractQueryFacets` match however the client
+   phrases it. `extractQueryFacets` already extracts an explicit number and derives one from
+   target + item type; the paired prompts should exercise both spellings.
+
+## The store and the verification gate (implemented)
+
+The paired store and the gate now exist:
+
+- **Store:** `spec/examples/<id>.gc` (the program, authored with `task-model tmN`) + `<id>.expect.json`
+  (`{ prompt, expect: { target, taskModel/itemType, dimension, standard } }`). First anchor in:
+  `c1-t9-tm3-ebsr` (the exact drift case). This is the human-refined artifact — author/edit the `.gc`.
+- **Gate:** `verifyExample({ errors, data, code, expect })` (exported from `@graffiticode/l0175`) —
+  BLOCKING checks (compiles clean; composed `target`/item type/`task-model:<n>`/dimension/standard
+  match the declared intent, read from `targets.ts` NOT `instructions.md`) + ADVISORY (compiler
+  warnings surfaced, not fatal). It never maps prompt→code; it validates whatever code exists, so it
+  guards **both** bootstrapped and hand-refined examples. Run it in the console capture step and after
+  any refinement.
+- **Corpus check:** `test/corpus.spec.ts` runs the gate over every `spec/examples/*.gc` on each test
+  run; `test/verify-example.spec.ts` proves the gate rejects a mislabeled example that compiles clean.
+
+**Coverage: 17 of 17 matrix cells — the full matrix.** Every `targets.<id>.taskModels` cell has a
+gated golden pair (all advisory-free): the c1-t4 / c1-t11 R&E trios, the c1-t1 / c1-t8 / c1-t10 trios, the c1-t2 five, and
+the complete c1-t9 five-item set (tm1 MC, tm2 multi-select, tm3 ebsr, tm4 hot-text, tm5 short-text).
+Each is a contrastive sibling set (same passage per target, varying only the task model). Any new or
+edited anchor is re-validated by the gate (`test/corpus.spec.ts`).

package/spec/scope.json

@@ -1,24 +1,25 @@
 {
-  "revised": "2026-06-24",
+  "revised": "2026-08-28",
   "id": "0175",
-  "summary": "Composes 5th-grade ELA assessment items (Smarter Balanced · Grade 5 · Claim 1) for multiple learning targets selected by a top-level `target`: c1-t4 (Reasoning & Evidence, literary, RL), c1-t11 (Reasoning & Evidence, informational, RI), c1-t9 (Central Ideas, informational, RI-1/RI-2), c1-t8 (Key Details, informational, RI-1/RI-7 — the inference is given and the student selects supporting evidence), and c1-t10 (Word Meanings, informational, RI-4/L-4 — the meaning of a targeted word, options authored as word/meaning) — different reading skills. Item-first: each authored question pins its correct answer(s) (focus) and its stem, and the compiler draws the wrong answers from the foils (distractor claims, non-supporting sources, or distractor meanings) bound to it.",
+  "summary": "Composes 5th-grade ELA assessment items (Smarter Balanced · Grade 5 · Claim 1) for multiple learning targets selected by a top-level `target`: c1-t4 (Reasoning & Evidence, literary, RL), c1-t11 (Reasoning & Evidence, informational, RI), c1-t2 (Central Ideas, literary, RL-1/RL-2 — theme, and summary scoped to a key event) and c1-t9 (Central Ideas, informational, RI-1/RI-2), c1-t1 (Key Details, literary, RL-1 alone) and c1-t8 (Key Details, informational, RI-1/RI-7) — for both, the inference is given and the student selects supporting evidence, and c1-t10 (Word Meanings, informational, RI-4/L-4 — the meaning of a targeted word, options authored as word/meaning) — different reading skills. Item-first: each authored question pins its correct answer(s) (focus) and its stem, and the compiler draws the wrong answers from the foils (distractor claims, non-supporting sources, or distractor meanings) bound to it.",
   "in_scope": [
-    "A required top-level target: c1-t4 (literary, R&E), c1-t11 (informational, R&E), c1-t9 (informational, Central Ideas), c1-t8 (informational, Key Details), or c1-t10 (informational, Word Meanings), selecting the dimension/standard/error-type/DOK/item-type/stem vocabulary",
+    "A required top-level target: c1-t4 (literary, R&E), c1-t11 (informational, R&E), c1-t9 (informational, Central Ideas), c1-t2 (literary, Central Ideas), c1-t1 (literary, Key Details), c1-t8 (informational, Key Details), or c1-t10 (informational, Word Meanings), selecting the dimension/standard/error-type/DOK/item-type/stem vocabulary",
     "A single passage with numbered lines",
     "Questions (outcomes) authored first, each with a unique id, a focus correct claim (a list on multi-select), and a stem from the target's guideline catalog; distractors bound to questions via targets",
     "A content graph: supported and distractor claims, plus evidence sources tagged directly-supports / supports-wrong-claim / irrelevant",
     "c1-t4 dimensions: character, setting, event, point-of-view, theme, topic, narrators-feelings, character-relationship",
     "c1-t11 dimensions: relationships-interactions, author-use-of-information, point-of-view, purpose, authors-opinion",
     "c1-t9 (Central Ideas) dimensions: central-idea, key-detail, summary",
-    "c1-t8 (Key Details) dimension: supporting-evidence — the inference is given in the stem and the student selects supporting evidence (options are sources, not claims)",
+    "c1-t2 (Central Ideas, literary) dimensions: theme, central-idea, key-detail, summary — theme is the lesson/message of a story; summary items cover a section or key event, never the whole text",
+    "c1-t1 (literary) / c1-t8 (informational) Key Details dimension: supporting-evidence — the inference is given in the stem and the student selects supporting evidence (options are sources, not claims)",
     "c1-t10 (Word Meanings) dimension: word-meaning — the question asks the meaning of a targeted word/phrase; options are meanings authored as word/meaning (not claims)",
     "Item types: ebsr (two-part selected response), hot-text (select text — sentence-level for R&E/Central-Ideas/Key Details, word-level click-the-word for Word Meanings), short-text (constructed response), multiple-choice (one correct), multi-select (exact correct set); the allowed set is per-target",
-    "Authored, error-typed distractors — R&E: misreads-detail / erroneous-inference / faulty-reasoning; Central Ideas (c1-t9): too-narrow / too-broad / misreads-detail / insignificant; Key Details (c1-t8): non-supporting sources (supports-wrong-claim / irrelevant), not distractor claims; Word Meanings (c1-t10): distractor meanings other-meaning / misinterprets / wrong-context — with rationales",
-    "Standards rl-1/rl-3/rl-6/rl-9 (c1-t4), ri-1/ri-3/ri-6/ri-7/ri-8/ri-9 (c1-t11), ri-1/ri-2 (c1-t9), ri-1/ri-7 (c1-t8), ri-4/l-4/l-4a/l-4b/l-4c/l-5c (c1-t10); DOK r-dok1..r-dok3 (R&E r-dok3; T9 r-dok2, written summary r-dok3; T8 & T10 r-dok2)",
-    "A reading-level target set by the guideline/target's grade (Grade 5 for c1-t4/c1-t11), overridable by an optional top-level `grade`; the compiler estimates the passage's reading level and warns when it runs above the target grade"
+    "Authored, error-typed distractors — R&E: misreads-detail / erroneous-inference / faulty-reasoning; Central Ideas (c1-t2 / c1-t9): too-narrow / too-broad / misreads-detail / insignificant; Key Details (c1-t1 / c1-t8): non-supporting sources (supports-wrong-claim / irrelevant), not distractor claims; Word Meanings (c1-t10): distractor meanings other-meaning / misinterprets / wrong-context — with rationales",
+    "Standards: the full CCSS Grade-5 strand for the target's text type is accepted (literary c1-t4/c1-t2/c1-t1 → RL rl-1..rl-7/rl-9, no rl-8; informational c1-t11/t9/t8/t10 → RI ri-1..ri-9; c1-t10 also the L-4/L-5 vocabulary families). The dimension's companion is inferred when standard is omitted — primary companions: c1-t4 rl-1 + rl-2 (theme/topic) / rl-3 / rl-6; c1-t11 ri-1 + ri-3/ri-6/ri-7/ri-8; c1-t9 ri-1+ri-2; c1-t2 rl-1+rl-2 (every dimension); c1-t1 rl-1 alone (no companion); c1-t8 ri-1+ri-7; c1-t10 ri-4 + the L-4 family. DOK r-dok1..r-dok3 (R&E r-dok3; T2/T9 r-dok2, the written response r-dok3; T1, T8 & T10 r-dok2)",
+    "A reading-level target set by the guideline/target's grade (Grade 5 for every current target), overridable by an optional top-level `grade`; the compiler estimates the passage's reading level and warns when it runs above the target grade"
   ],
   "out_of_scope": [
-    "Claims, grades, or Claim-1 targets beyond T4, T11, T9, T8, and T10 — other Claim-1 targets belong in their own dialects",
+    "Claims, grades, or Claim-1 targets beyond T4, T11, T2, T9, T1, T8, and T10 — other Claim-1 targets belong in their own dialects",
     "Multi-word-phrase candidates in c1-t10 click-the-word (single words only this version)",
     "Dual-text stimuli (a single passage only in this version)",
     "Compile-time LLM generation — the compiler selects and validates authored content, it does not invent it",

package/spec/spec.md

@@ -7,7 +7,9 @@ _Revised: 2026-06-19_
 items (Smarter Balanced · Grade 5 · Claim 1). One language serves **multiple learning targets**,
 selected by a required top-level `target`: `c1-t4` (Reasoning & Evidence, literary, RL standards),
 `c1-t11` (Reasoning & Evidence, informational, RI standards), `c1-t9` (Central Ideas,
-informational, RI-1/RI-2), `c1-t8` (Key Details, informational, RI-1/RI-7 — the inference is
+informational, RI-1/RI-2), `c1-t2` (Central Ideas, **literary**, RL-1/RL-2 — theme, and summary scoped to a key event),
+`c1-t1` (Key Details, **literary**, RL-1 — the inference is given and the student selects
+supporting evidence), `c1-t8` (Key Details, informational, RI-1/RI-7 — the inference is
 given and the student selects supporting evidence), or `c1-t10` (Word Meanings, informational,
 RI-4/L-4 — the meaning of a targeted word, options authored as `word`/`meaning`) — different
 reading skills, each with its own dimensions, distractor taxonomy, DOK, item types, and stem
@@ -64,7 +66,7 @@ Free text (`text`, `rationale`, `subject`, `stem`, the passage heading) and id l
 
 | Form | Arity | Takes | Description |
 | :--- | :---: | :--- | :--- |
-| `target` | 2 | tag | Top level: the learning target — `c1-t4` (R&E literary, RL), `c1-t11` (R&E informational, RI), `c1-t9` (Central Ideas informational, RI-1/RI-2), `c1-t8` (Key Details informational, RI-1/RI-7), or `c1-t10` (Word Meanings informational, RI-4/L-4). Selects the valid dimensions/standards, distractor taxonomy, DOK, item types, and stem catalog. Always author one; defaults to `c1-t4` if omitted. |
+| `target` | 2 | tag | Top level: the learning target — `c1-t4` (R&E literary, RL), `c1-t11` (R&E informational, RI), `c1-t9` (Central Ideas informational, RI-1/RI-2), `c1-t2` (Central Ideas literary, RL-1/RL-2), `c1-t1` (Key Details literary, RL-1 alone), `c1-t8` (Key Details informational, RI-1/RI-7), or `c1-t10` (Word Meanings informational, RI-4/L-4). Selects the valid dimensions/standards, distractor taxonomy, DOK, item types, and stem catalog. Always author one; defaults to `c1-t4` if omitted. |
 | `title` | 2 | string | Optional assessment title; echoed on the composed output. |
 | `grade` | 2 | number | Optional top-level reading-level target (e.g. `grade 5`). Defaults to the target/guideline's grade (5 for `c1-t4`/`c1-t11`); echoed on the output. The compiler estimates the passage's reading level and warns when it reads above this grade. |
 | `passage` | 2 | string | Opens the stimulus; the value is the passage **heading**. Chains with `type` and `lines`. |
@@ -144,14 +146,14 @@ composes). A `focus` that isn't a supported claim, or a `targets` to a missing o
 
 ## Enumerations
 
-- **`target`**: `c1-t4` (R&E literary), `c1-t11` (R&E informational), `c1-t9` (Central Ideas informational), `c1-t8` (Key Details informational), `c1-t10` (Word Meanings informational) — required, top level
+- **`target`**: `c1-t4` (R&E literary), `c1-t11` (R&E informational), `c1-t9` (Central Ideas informational), `c1-t2` (Central Ideas **literary**), `c1-t1` (Key Details **literary**), `c1-t8` (Key Details informational), `c1-t10` (Word Meanings informational) — required, top level
 - **item `type`**: `ebsr`, `hot-text`, `short-text`, `multiple-choice`, `multi-select` (allowed set is per-target) · **passage `type`**: `literary`, `informational`
 - **`dimension` (c1-t4)**: `character`, `setting`, `event`, `point-of-view`, `theme`, `topic`, `narrators-feelings`, `character-relationship`
 - **`dimension` (c1-t11)**: `relationships-interactions`, `author-use-of-information`, `point-of-view`, `purpose`, `authors-opinion`
-- **`dimension` (c1-t9)**: `central-idea`, `key-detail`, `summary` · **(c1-t8)**: `supporting-evidence` · **(c1-t10)**: `word-meaning`
+- **`dimension` (c1-t2)**: `theme`, `central-idea`, `key-detail`, `summary` · **(c1-t9)**: `central-idea`, `key-detail`, `summary` · **(c1-t1 / c1-t8)**: `supporting-evidence` · **(c1-t10)**: `word-meaning`
 - **claim `status`**: `supported`, `distractor` · **source `status`**: `directly-supports`, `supports-wrong-claim`, `irrelevant` · **meaning `status` (c1-t10)**: `correct`, `distractor`
-- **`error-type` (c1-t4 / c1-t11)**: `misreads-detail`, `erroneous-inference`, `faulty-reasoning` · **(c1-t9)**: `too-narrow`, `too-broad`, `misreads-detail`, `insignificant` · **(c1-t8)**: none — non-supporting sources · **(c1-t10)**: `other-meaning`, `misinterprets`, `wrong-context`
-- **`standard` (c1-t4)**: `rl-1`, `rl-3`, `rl-6`, `rl-9` · **(c1-t11)**: `ri-1`, `ri-3`, `ri-6`, `ri-7`, `ri-8`, `ri-9` · **(c1-t9)**: `ri-1`, `ri-2` · **(c1-t8)**: `ri-1`, `ri-7` · **(c1-t10)**: `ri-4`, `l-4`, `l-4a`, `l-4b`, `l-4c`, `l-5c` · **`dok`**: `r-dok1`, `r-dok2`, `r-dok3`
+- **`error-type` (c1-t4 / c1-t11)**: `misreads-detail`, `erroneous-inference`, `faulty-reasoning` · **(c1-t2 / c1-t9)**: `too-narrow`, `too-broad`, `misreads-detail`, `insignificant` · **(c1-t1 / c1-t8)**: none — non-supporting sources · **(c1-t10)**: `other-meaning`, `misinterprets`, `wrong-context`
+- **`standard`** — the full CCSS Grade-5 strand for the target's text type is accepted; the dimension's companion is inferred when you omit it. Primary companions: **(c1-t4)** `rl-1` + `rl-2` (theme/topic) / `rl-3` / `rl-6` — full **RL** strand `rl-1`–`rl-7`/`rl-9` accepted (no `rl-8`) · **(c1-t11)** `ri-1` + `ri-3` / `ri-6` / `ri-7` / `ri-8` · **(c1-t2)** `rl-1` + `rl-2` · **(c1-t9)** `ri-1` + `ri-2` · **(c1-t1)** `rl-1` **alone** — no companion in the guideline · **(c1-t8)** `ri-1` + `ri-7` — c1-t11/t9/t8 accept the full **RI** strand `ri-1`–`ri-9` · **(c1-t10)** `ri-4` + the `l-4` / `l-5` families. **`dok`**: `r-dok1`, `r-dok2`, `r-dok3`
 
 ## How composition uses the vocabulary
 
@@ -186,6 +188,11 @@ lines [
   "Mara crouched at the edge of the tide pool, ignoring the picnic behind her."
   "Her brother called twice, but she did not turn around."
   "A tiny crab scuttled under a rock, and Mara smiled for the first time all day."
+  "She traced the cold water as if the pool were the only thing that mattered."
+  "Behind her, paper plates rustled and her mother laughed."
+  "Someone asked whether she wanted a sandwich, and she said nothing at all."
+  "Only when her father folded the last chair did Mara stand up."
+  "The tide crept in and filled the pool to its rim."
 ]
 claims [
   claim id "c1"
@@ -193,35 +200,58 @@ claims [
     dimension character
     subject "Mara"
     standard rl-1
-    text "Mara is more interested in the tide pool than in her family's picnic."
+    text "Mara cares more about the tide pool than about the picnic."
     cites [
       "e1"
       "e3"
+      "e4"
     ] {}
   claim id "c2"
     status distractor
     error-type misreads-detail
     plausibility 0.8
     targets [ "q1" ]
-    text "Mara is angry at her brother."
+    text "Mara is angry at her brother for calling her twice."
     rationale "Not turning around shows absorption, not anger."
     cites [
       "e2"
     ] {}
   claim id "c3"
     status distractor
-    error-type erroneous-inference
+    error-type misreads-detail
+    plausibility 0.6
     targets [ "q1" ]
-    text "Mara dislikes the outdoors."
-    rationale "Contradicted by her smile in line 3."
+    text "Mara is bored by the pool and wants to go home."
+    rationale "Her stillness is focus, not boredom."
     cites [
       "e2"
     ] {}
   claim id "c4"
+    status distractor
+    error-type erroneous-inference
+    plausibility 0.55
+    targets [ "q1" ]
+    text "Mara would rather be indoors than out at the beach."
+    rationale "Contradicted by her smile at the crab."
+    cites [
+      "e3"
+    ] {}
+  claim id "c5"
+    status distractor
+    error-type erroneous-inference
+    plausibility 0.5
+    targets [ "q1" ]
+    text "Mara is waiting for her brother to come look with her."
+    rationale "Invents a goal the passage never states."
+    cites [
+      "e2"
+    ] {}
+  claim id "c6"
     status distractor
     error-type faulty-reasoning
+    plausibility 0.45
     targets [ "q1" ]
-    text "Because Mara is quiet, she must be upset."
+    text "Mara is quiet, so something must have upset her."
     rationale "Treats quiet as upset without support."
     cites [
       "e2"
@@ -230,27 +260,63 @@ claims [
 evidence [
   source id "e1"
     line 1
+    quote "Mara crouched at the edge of the tide pool, ignoring the picnic behind her."
     status directly-supports
     supports [
       "c1"
     ] {}
+  source id "e3"
+    line 3
+    quote "A tiny crab scuttled under a rock, and Mara smiled for the first time all day."
+    status directly-supports
+    supports [
+      "c1"
+    ] {}
+  source id "e4"
+    line 4
+    quote "She traced the cold water as if the pool were the only thing that mattered."
+    status directly-supports
+    supports [
+      "c1"
+    ] {}
+  /* NO-GIVEAWAY: a supports-wrong-claim source backing BOTH the key and a distractor, so
+     Part B does not telegraph Part A */
   source id "e2"
     line 2
+    quote "Her brother called twice, but she did not turn around."
     status supports-wrong-claim
     supports [
       "c1"
       "c2"
     ] {}
-  source id "e3"
-    line 3
-    status directly-supports
+  source id "e6"
+    line 6
+    quote "Someone asked whether she wanted a sandwich, and she said nothing at all."
+    status supports-wrong-claim
     supports [
       "c1"
+      "c6"
     ] {}
+  source id "e5"
+    line 5
+    quote "Behind her, paper plates rustled and her mother laughed."
+    status irrelevant
+    supports [] {}
+  source id "e7"
+    line 7
+    quote "Only when her father folded the last chair did Mara stand up."
+    status irrelevant
+    supports [] {}
+  source id "e8"
+    line 8
+    quote "The tide crept in and filled the pool to its rim."
+    status irrelevant
+    supports [] {}
 ]
 outcomes [
   outcome id "q1"
     type ebsr
+    task-model tm1
     dimension character
     subject "Mara"
     standard rl-1
@@ -259,6 +325,7 @@ outcomes [
     stem-b "Which sentence(s) from the passage best support your answer in Part A?" {}
   outcome id "q2"
     type short-text
+    task-model tm3
     dimension character
     subject "Mara"
     standard rl-1
@@ -275,3 +342,11 @@ outcomes [
 ]
 {}..
 ```
+
+The example above is Target 4 (literary, R&E). The other targets keep the same flat-chain shape
+but differ in what the options are: **T11** is the same R&E shape over an informational passage
+(RI standards); **T9** (Central Ideas) options are `claim`s judged by significance (`too-narrow` /
+`too-broad` / `insignificant` / `misreads-detail`); **T8** (Key Details) gives the inference in the
+stem and the options are `source`s (no distractor claims); **T10** (Word Meanings) options are
+`meaning`s of a targeted `word` authored in a top-level `words` list. See `instructions.md` for a
+full worked program per target and `stems.md` for each target's stem catalog.

package/spec/stems.md

@@ -1,7 +1,7 @@
 <!-- SPDX-License-Identifier: CC-BY-4.0 -->
 # L0175 Appropriate-Stem Catalog (SBAC · Grade 5 · Claim 1)
 
-_Revised: 2026-06-24_
+_Revised: 2026-08-28_
 
 The **code generator authors each question's stem from this catalog** and emits it on the
 outcome (`stem`, and `stem-b` on EBSR). The compiler uses the authored text verbatim — it does
@@ -12,7 +12,9 @@ dual-text-stimuli stems are out of scope).
 **First match your program's `target`, then use that target's catalog below:**
 - **`c1-t4`** (Reasoning & Evidence, literary) → `packages/core/data/E.G5.C1.T4 Reasoning & Evidence.pdf`
 - **`c1-t11`** (Reasoning & Evidence, informational) → `packages/core/data/E.G5.C1.T11 Reasoning & Evidence.pdf`
+- **`c1-t2`** (Central Ideas, **literary**) → `packages/core/data/E.G5.C1.T2 Central Ideas.pdf`
 - **`c1-t9`** (Central Ideas, informational) → `packages/core/data/E.G5.C1.T9 Central Ideas.pdf`
+- **`c1-t1`** (Key Details, **literary**) → `packages/core/data/E.G5.C1.T1 Key Details.pdf`
 - **`c1-t8`** (Key Details, informational) → `packages/core/data/E.G5.C1.T8 Key Details.pdf`
 - **`c1-t10`** (Word Meanings, informational) → `packages/core/data/E.G5.C1.T10 Word Meanings.pdf`
 
@@ -21,13 +23,29 @@ dual-text-stimuli stems are out of scope).
 1. Use the section for your `target`.
 2. Pick the stem that matches your **item `type`** and the task. Task-model numbering is
    per-target and differs between targets — match by item type, not by number. (R&E targets:
-   EBSR=TM1, Hot-Text=TM2, Short-Text=TM3. T9 Central Ideas: Multiple-Choice=TM1,
-   Multi-Select=TM2, EBSR=TM3, Hot-Text=TM4, Short-Text=TM5. T8 Key Details: Multiple-Choice=TM1,
-   Multi-Select=TM2, Hot-Text=TM3. T10 Word Meanings: Multiple-Choice=TM1, Multi-Select=TM2,
-   Hot-Text=TM3.)
+   EBSR=TM1, Hot-Text=TM2, Short-Text=TM3. T2 and T9 Central Ideas: Multiple-Choice=TM1,
+   Multi-Select=TM2, EBSR=TM3, Hot-Text=TM4, Short-Text=TM5. T1 and T8 Key Details:
+   Multiple-Choice=TM1, Multi-Select=TM2, Hot-Text=TM3. T10 Word Meanings: Multiple-Choice=TM1,
+   Multi-Select=TM2, Hot-Text=TM3.)
 3. Pick the **one stem template** that matches the task and the dimension, and fill its
    bracketed `[...]` slot.
 
+Task-model number → item type, per target (the numbers collide across targets, so a number is
+meaningless without its target):
+
+<!-- GENERATED:task-models START (from targets.json — regenerated by tools/build-static.js; do not edit by hand) -->
+| Target | tm1 | tm2 | tm3 | tm4 | tm5 |
+|--------|-----|-----|-----|-----|-----|
+| `c1-t4` — Grade 5 · Claim 1 · Target 4 (Reasoning & Evidence) | ebsr | hot-text | short-text | — | — |
+| `c1-t11` — Grade 5 · Claim 1 · Target 11 (Reasoning & Evidence) | ebsr | hot-text | short-text | — | — |
+| `c1-t2` — Grade 5 · Claim 1 · Target 2 (Central Ideas) | multiple-choice | multi-select | ebsr | hot-text | short-text |
+| `c1-t9` — Grade 5 · Claim 1 · Target 9 (Central Ideas) | multiple-choice | multi-select | ebsr | hot-text | short-text |
+| `c1-t1` — Grade 5 · Claim 1 · Target 1 (Key Details) | multiple-choice | multi-select | hot-text | — | — |
+| `c1-t8` — Grade 5 · Claim 1 · Target 8 (Key Details) | multiple-choice | multi-select | hot-text | — | — |
+| `c1-t10` — Grade 5 · Claim 1 · Target 10 (Word Meanings) | multiple-choice | multi-select | hot-text | — | — |
+<!-- GENERATED:task-models END -->
+
+
 **Specificity rule (required, both targets).** The `[...]` slot must name the concrete thing the
 question is about — a character's name (`Mother`), a specific event (`the turkey-feeding`), a
 specific idea (`the relationship between the bridge designs`), the author's point of view, etc.
@@ -111,6 +129,15 @@ Author as `stem` (the prompt); every Short Text stem ends with the explain claus
   `stem "What did the author most likely mean by including the scarecrow in the passage?"`
 - theme / inference / Short Text →
   `stem "What inference can be made about the theme? Explain using key details from the passage to support your answer."`
+- topic / Short Text →
+  `stem "What is the story mostly about? Explain using key details from the passage to support your answer."`
+
+**`theme` vs `topic` (both → `rl-2`, but they are different dimensions).** `topic` is what the
+text is *about* (its subject) — cued by "**mostly about**" / "**what is the passage/story about**".
+`theme` is the *lesson or message* the text conveys — cued by "**the theme**" / "**the message**"
+/ "**the lesson**". Route a "mostly about" prompt to `topic`, and write the correct answer as a
+subject statement ("The story is mostly about …"), **not** a life-lesson statement. Reserve the
+lesson-style answer for `theme`.
 
 ---
 
@@ -124,6 +151,14 @@ concepts / author's use of information / point of view / purpose]` (or, for an o
 "key **evidence**" (T4 said "key details"), and Part B units are
 `sentence(s) / paragraph(s) / section(s)`.
 
+**`purpose` vs `point-of-view` (both start "author's …" but are different dimensions with
+different companions).** `purpose` = *why the author wrote the text* / what they are trying to do
+(inform, persuade, explain) — cued by "**author's purpose**" / "why the author wrote" → companion
+`ri-8`. `point-of-view` = *the author's stance or attitude* toward the subject — cued by
+"**author's point of view**" / "how the author feels about" → companion `ri-6`. Route a
+"purpose" prompt to `purpose`/`ri-8` and a "point of view" prompt to `point-of-view`/`ri-6`;
+easiest is to **omit `standard`** and let the dimension infer its companion.
+
 ## Task Model 1 — EBSR (two-part selected response)
 
 **Lead-in:** This question has two parts. First, answer part A. Then, answer part B.
@@ -186,6 +221,90 @@ Author as `stem`; every Short Text stem ends with the explain clause ("key **evi
 
 ---
 
+# Target 2 (`c1-t2`) — Central Ideas (**literary**)
+
+The **literary twin of Target 9**, and the only other target that offers **all five item types**.
+DOK **2** (**3** for the written response). Standards **RL-1 + RL-2**. Dimensions: `theme`,
+`central-idea`, `key-detail`, `summary` — every one answers to `rl-2`, so a composed item's
+`standards` is `["rl-1", "rl-2"]`. Distractors use the same **significance** taxonomy as T9
+(`too-narrow`, `too-broad`, `misreads-detail`, `insignificant`) — usually *true* statements that
+simply aren't central.
+
+**Three things T2 does that T9 does not:**
+
+1. **Theme leads.** RL-2 is "determine a theme of a story, drama, or poem from details in the
+   text." Prefer the `theme` dimension for a story or poem and phrase the correct answer as a
+   *lesson or message* ("Keeping at something can matter more than winning"), not a plot summary.
+   Use `central-idea` when the prompt says "main idea" rather than "theme" or "message".
+2. **Summary is SCOPED.** The guideline is explicit: *items will not ask students to summarize the
+   entire text; students will summarize a key event(s) or idea(s).* Point every `summary` stem at a
+   section or a key event ("Summarize what happens after…", "…the first paragraph"). A
+   "summarize the passage" stem is out of scope on T2 — that phrasing belongs to T9.
+3. **`key-detail` covers key EVENTS.** In a story the details that build a theme are usually
+   events, so tag them `key-detail`.
+
+**A caution from the guideline, for `insignificant`:** distractors "will not contain details of such
+insignificance that the options are implausible." A foil so trivial that no student would pick it
+is a wasted option — keep it true, relevant, and merely *not central*. Likewise, TM1 warns that
+distractors must not turn on sequencing or other DOK-1 errors.
+
+## Task Model 1 — Multiple Choice (`type multiple-choice`, single-answer, DOK 2)
+
+- Which sentence best [shows/tells/describes] the [theme/main idea] of the passage?
+- Which sentence best [shows/tells/describes] the [theme/main idea] of the [story/poem] told by the [narrator/speaker]?
+- Which sentence best [shows/tells/describes] [the author's/character's] message about [provide theme/main idea]?
+- Which [sentences/lines/paragraph(s)] from the passage best [shows/tells/describes] the [theme/main idea]?
+- Which sentence best summarizes the [first paragraph/introductory paragraph]?
+- Which sentence best summarizes what happens after [provide plot]?
+- Read this summary. [summary of a section; one key detail/event is missing] Which [key detail/event] is missing from the summary? *(the correct answer must be a pivotal key event or idea)*
+
+## Task Model 2 — Multi-Select (`type multi-select`, choose two, DOK 2)
+
+`focus` is a **list** of the two correct claims.
+- Choose two sentences that best [show/tell/describe] the [theme(s)/main idea(s)] presented in the passage.
+- Choose two sentences that should be included in a summary of [the text/section of the text].
+
+## Task Model 3 — EBSR (`type ebsr`, two-part, DOK 2)
+
+Lead-in: *The following question has two parts. First, answer part A. Then, answer part B.*
+Part A:
+- Which sentence best [shows/tells/describes] the [theme/main idea] of the text?
+- Which sentence best [shows/tells/describes] the [author's/character's] [message/main idea]?
+
+Part B (`stem-b`):
+- Which [sentence(s)/line(s)/paragraph(s)] from the passage best [support(s)/is (are) an example of] your answer in part A?
+
+**No-giveaway (the guideline states it outright):** avoid Part A/Part B choices that obviously
+correspond, so that picking a Part A option determines the Part B option. Some or all Part B options
+should plausibly support **more than one** Part A choice — author at least one
+`supports-wrong-claim` source listing BOTH the correct claim and a distractor.
+
+## Task Model 4 — Hot Text (`type hot-text`, single-part select-text, DOK 2)
+
+Single-part: the authored `stem` is the whole instruction; the student clicks the sentence(s) whose
+`source`s `directly-supports` the focus claim. The stem states how many to click.
+- Click on [one/two] [sentence(s)/line(s)/set of lines/paragraph(s)] that best [shows/tells/describes] the [theme/main idea] of [the text/section of the text]. Choose [one/two] answer(s).
+- Read this summary of [the text/section of the text]. [summary; one key event missing] A [key event/detail] is missing from the summary. Click on the [sentence(s)/paragraph(s)] that include(s) the missing [event/detail].
+- Click on the [one/two] [key detail(s)/event(s)] that should be included in a summary of the text.
+
+## Task Model 5 — Short Text (`type short-text`, constructed response, DOK 3)
+
+Written response; **never the whole text** (see #2 above).
+- Determine the [theme/main idea] of the passage. Explain the [theme/main idea] using key details from the passage to support your answer.
+- What is the [theme/main idea] of the passage? Explain using key details from the passage to support your answer.
+- Summarize the [first paragraph/introductory paragraph]. Use key [details/events] from the passage in your summary.
+- Summarize what happens [after/during] [provide text event]. Use key [details/events] from the passage in your summary.
+
+## Worked examples (T2, specific slot fills)
+
+- theme / Multiple Choice → `stem "Which sentence best tells the theme of the passage?"` with the correct claim phrased as a lesson ("Keeping at something can matter more than winning.").
+- summary / Multi-Select → `stem "Choose two sentences that should be included in a summary of the story. Select two answers."` with `focus ["s1" "s2"]`.
+- theme / EBSR → Part A as above, `stem-b "Which sentence from the passage best supports your answer in Part A?"`.
+- theme / Hot Text → `stem "Click on the two sentences that best show the theme of the story. Choose two answers."` (author exactly two `directly-supports` sources).
+- summary / Short Text → `stem "Summarize what happens after the blue ribbon goes to somebody else. Use key details from the passage in your summary."` — scoped to a key event, not the whole story.
+
+---
+
 # Target 9 (`c1-t9`) — Central Ideas (informational)
 
 A **different skill** from Reasoning & Evidence: synthesize and condense — the **main/central
@@ -264,6 +383,63 @@ it); the student clicks the sentence(s) that show the main idea. The correct sen
 
 ---
 
+# Target 1 (`c1-t1`) — Key Details (**literary**)
+
+The **literary mirror of Target 8**: the inference/conclusion is **GIVEN in the stem**, and the
+student selects the **text evidence** that supports it. DOK **1–2**. Standard **RL-1 — and only
+RL-1** (unlike T8 the guideline names no companion, so a composed item's `standards` is just
+`rl-1`; omit `standard` and let the dimension resolve it). Dimension: `supporting-evidence`. Item
+types: `multiple-choice`, `multi-select`, `hot-text` (no EBSR, no constructed response).
+
+**Authoring:** identical to T8 — author ONE supported `claim` = the given inference (named by the
+outcome's `focus`) and **state that inference in the `stem`**; author `source`s for the options —
+`directly-supports` (the correct evidence; give a `quote` for the exact sentence) and
+`supports-wrong-claim` / `irrelevant` (the distractor evidence, each with a `rationale`). The
+options are the **sources**, not claims; there are no distractor claims.
+
+**Two literary differences from T8's wording:** these stems offer **`line`** as a selectable unit
+(matching L0175's numbered `lines`) and say **`[author/narrator]`** where T8 says only "author" —
+use "narrator" for a first-person story. Multi-Select is exactly **two** correct (T8 allows two or
+three).
+
+## Task Model 1 — Multiple Choice (`type multiple-choice`, single-answer, DOK 1–2)
+
+The stem states the inference and asks which evidence supports it:
+- Which [detail/sentence/line] from the passage best [supports/shows] [provide inference or conclusion based on the passage]?
+- [Provide inference or conclusion based on the passage]. Which [detail/sentence/line] from the passage best supports this [inference/conclusion] OR best shows [provide the inference or conclusion]?
+- The reader can [infer/conclude] [inference or conclusion based on the passage]. Which [detail/sentence/line] from the passage best supports this [inference/conclusion]?
+- The [author/narrator] [infers/concludes] that [inference/conclusion based on the passage]. Which [detail/sentence/line] from the passage best supports this [inference/conclusion]?
+- Read this [inference/conclusion]. [Provide inference or conclusion based on the passage]. Which [detail/sentence/line] from the passage best supports this [inference/conclusion]?
+
+## Task Model 2 — Multi-Select (`type multi-select`, choose two, DOK 1–2)
+
+- Choose two [details/sentences/lines] from the passage that best support the [inference/conclusion] that [provide inference or conclusion based on the passage].
+- Which [details/sentences/lines] from the passage best support [provide inference or conclusion based on the passage]? Select two answers.
+- [Provide inference or conclusion based on the passage]. Which [details/sentences/lines] from the passage best support this [inference/conclusion]? Select two answers.
+- The reader can [infer/conclude] [provide inference/conclusion based on the passage]. Which [details/sentences/lines] from the passage best support this [inference/conclusion]? Select two answers.
+- The [author/narrator] [infers/concludes] that [provide inference/conclusion based on the passage]. Which [details/sentences/lines] from the passage best support this [inference/conclusion]? Select two answers.
+
+## Task Model 3 — Hot Text (`type hot-text`, single-part select-text, DOK 1–2)
+
+Single-part: the authored `stem` states the inference and the click instruction (the compiler does
+not synthesize it); the student clicks the supporting sentence(s):
+- Click the [detail/sentence/set of sentences/line/paragraph] that best supports [provide inference or conclusion based on the passage].
+- Click [one/two] [details/sentences/sets of sentences/lines/paragraphs] that best support [provide inference or conclusion based on the passage].
+- [Provide inference or conclusion based on the passage]. Click the [detail/sentence/set of sentences/line/paragraph] that best supports this [inference/conclusion].
+- The reader can [infer/conclude] [provide inference/conclusion based on the passage]. Click [one/two] [details/sentences/lines] that best support this [inference/conclusion].
+- Read this [inference/conclusion]. [Provide inference or conclusion based on the text]. Click [one/two] [details/sentences/sets of sentences/lines/paragraphs] that best support this [inference/conclusion].
+
+## Worked examples (T1, specific slot fills)
+
+- supporting-evidence / Multiple Choice →
+  `stem "The reader can conclude that Nina takes care of a problem on her own, without being told to. Which line from the passage best supports this conclusion?"` with one supported `claim` (the inference) at `focus`, its directly-supporting `source` the correct option.
+- supporting-evidence / Multi-Select →
+  `stem "Nina takes care of a problem on her own, without being told to. Which two lines from the passage best support this conclusion? Select two answers."` (two `directly-supports` sources)
+- supporting-evidence / Hot Text (single-part) →
+  `stem "Read this conclusion. Nina takes care of a problem on her own, without being told to. Click the two sentences from the passage that best support this conclusion."`
+
+---
+
 # Target 8 (`c1-t8`) — Key Details (informational)
 
 A **different model**: the inference/conclusion is **GIVEN in the stem**, and the student selects