zam-core 0.6.1 → 0.6.2

package/dist/cli/index.js

@@ -552,7 +552,9 @@ CREATE TABLE IF NOT EXISTS tokens (
   created_at    TEXT NOT NULL DEFAULT (datetime('now')),
   updated_at    TEXT NOT NULL DEFAULT (datetime('now')),
   deprecated_at TEXT,
-  question      TEXT
+  question      TEXT,
+  provider      TEXT,
+  topic_id      TEXT
 );
 
 -- Prerequisite dependency graph: "to learn A, first know B"
@@ -992,6 +994,14 @@ async function runMigrations(db) {
       PRIMARY KEY (token_id, source_id)
     )
   `);
+  if (tokenCols.length > 0) {
+    if (!tokenCols.some((c) => c.name === "provider")) {
+      await db.exec(`ALTER TABLE tokens ADD COLUMN provider TEXT`);
+    }
+    if (!tokenCols.some((c) => c.name === "topic_id")) {
+      await db.exec(`ALTER TABLE tokens ADD COLUMN topic_id TEXT`);
+    }
+  }
 }
 
 // src/kernel/db/snapshot.ts
@@ -1485,8 +1495,8 @@ async function createToken(db, input8) {
     throw new Error(`bloom_level must be between 1 and 5, got ${bloom}`);
   }
   await db.prepare(`
-    INSERT INTO tokens (id, slug, concept, domain, bloom_level, context, symbiosis_mode, source_link, question, created_at, updated_at)
-    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+    INSERT INTO tokens (id, slug, concept, domain, bloom_level, context, symbiosis_mode, source_link, question, provider, topic_id, created_at, updated_at)
+    VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
   `).run(
     id,
     input8.slug,
@@ -1497,16 +1507,33 @@ async function createToken(db, input8) {
     input8.symbiosis_mode ?? null,
     input8.source_link ?? null,
     input8.question ?? null,
+    input8.provider ?? null,
+    input8.topic_id ?? null,
     now,
     now
   );
   return await getTokenById(db, id);
 }
+function parseTokenFallback(token) {
+  if (token && !token.provider && token.source_link) {
+    if (token.source_link.includes("lehrplanplus.bayern.de")) {
+      token.provider = "lehrplanplus-bayern";
+      const match = token.source_link.match(/#(.*)$/);
+      if (match) {
+        token.topic_id = match[1];
+      }
+    }
+  }
+}
 async function getTokenBySlug(db, slug) {
-  return await db.prepare("SELECT * FROM tokens WHERE slug = ?").get(slug);
+  const token = await db.prepare("SELECT * FROM tokens WHERE slug = ?").get(slug);
+  parseTokenFallback(token);
+  return token;
 }
 async function getTokenById(db, id) {
-  return await db.prepare("SELECT * FROM tokens WHERE id = ?").get(id);
+  const token = await db.prepare("SELECT * FROM tokens WHERE id = ?").get(id);
+  parseTokenFallback(token);
+  return token;
 }
 async function updateToken(db, slug, updates) {
   const token = await getTokenBySlug(db, slug);
@@ -1552,6 +1579,14 @@ async function updateToken(db, slug, updates) {
     fields.push("question = ?");
     values.push(updates.question);
   }
+  if (updates.provider !== void 0) {
+    fields.push("provider = ?");
+    values.push(updates.provider);
+  }
+  if (updates.topic_id !== void 0) {
+    fields.push("topic_id = ?");
+    values.push(updates.topic_id);
+  }
   if (fields.length === 0) {
     throw new Error("updateToken called with no fields to update");
   }
@@ -1677,14 +1712,20 @@ async function findTokens(db, query) {
   return scored;
 }
 async function listTokens(db, options) {
+  let tokens;
   if (options?.domain) {
-    return await db.prepare(
+    tokens = await db.prepare(
       "SELECT * FROM tokens WHERE domain = ? AND deprecated_at IS NULL ORDER BY bloom_level, slug"
     ).all(options.domain);
+  } else {
+    tokens = await db.prepare(
+      "SELECT * FROM tokens WHERE deprecated_at IS NULL ORDER BY bloom_level, domain, slug"
+    ).all();
   }
-  return await db.prepare(
-    "SELECT * FROM tokens WHERE deprecated_at IS NULL ORDER BY bloom_level, domain, slug"
-  ).all();
+  for (const token of tokens) {
+    parseTokenFallback(token);
+  }
+  return tokens;
 }
 function slugify(text) {
   return text.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/(^-|-$)/g, "");
@@ -1736,6 +1777,8 @@ async function listPersonalCards(db, userId, options) {
       t.question,
       t.created_at AS createdAt,
       t.updated_at AS updatedAt,
+      t.provider,
+      t.topic_id AS topicId,
       c.id AS cardId,
       c.state,
       c.due_at AS dueAt,
@@ -1765,6 +1808,17 @@ async function listPersonalCards(db, userId, options) {
   }
   sql += " ORDER BY t.created_at DESC";
   const rows = await db.prepare(sql).all(...values);
+  for (const row of rows) {
+    if (!row.provider && row.sourceLink) {
+      if (row.sourceLink.includes("lehrplanplus.bayern.de")) {
+        row.provider = "lehrplanplus-bayern";
+        const match = row.sourceLink.match(/#(.*)$/);
+        if (match) {
+          row.topicId = match[1];
+        }
+      }
+    }
+  }
   return rows;
 }
 async function importCurriculumCards(db, userId, cards) {
@@ -1811,9 +1865,18 @@ async function importCurriculumCards(db, userId, cards) {
           context: card.context || "",
           symbiosis_mode: symbiosisMode,
           source_link: card.source_link || null,
-          question: card.question || null
+          question: card.question || null,
+          provider: card.provider || null,
+          topic_id: card.topic_id || null
         });
         createdCount++;
+      } else {
+        if (!token.provider && card.provider) {
+          await updateToken(tx, token.slug, {
+            provider: card.provider,
+            topic_id: card.topic_id || null
+          });
+        }
       }
       const existingCard = await getCard(tx, token.id, userId);
       if (!existingCard) {
@@ -2035,11 +2098,12 @@ async function confirmSourceImport(db, userId, sourceId, proposals) {
   let createdCount = 0;
   let linkedCount = 0;
   await db.transaction(async (tx) => {
-    const source = await tx.prepare("SELECT id FROM sources WHERE id = ?").get(sourceId);
-    if (!source) {
-      throw new Error(`Source not found: ${sourceId}`);
-    }
     for (const card of proposals) {
+      const cardSourceId = card.source_id || sourceId;
+      const source = await tx.prepare("SELECT id FROM sources WHERE id = ?").get(cardSourceId);
+      if (!source) {
+        throw new Error(`Source not found: ${cardSourceId}`);
+      }
       const baseText = card.question && card.question.trim().length > 0 ? card.question : card.concept;
       const cleanDomain = slugify(card.domain || "");
       const cleanBase = slugify(baseText);
@@ -2070,11 +2134,19 @@ async function confirmSourceImport(db, userId, sourceId, proposals) {
           bloom_level: bloom,
           context: card.excerpt || "",
           symbiosis_mode: symbiosisMode,
-          question: card.question || null
+          question: card.question || null,
+          provider: card.provider || null,
+          topic_id: card.topic_id || null
         });
         createdCount++;
       } else {
         linkedCount++;
+        if (!token.provider && card.provider) {
+          await updateToken(tx, token.slug, {
+            provider: card.provider,
+            topic_id: card.topic_id || null
+          });
+        }
       }
       const existingCard = await getCard(tx, token.id, userId);
       if (!existingCard) {
@@ -2086,7 +2158,12 @@ async function confirmSourceImport(db, userId, sourceId, proposals) {
            ON CONFLICT(token_id, source_id) DO UPDATE SET
              excerpt = excluded.excerpt,
              page_number = excluded.page_number`
-      ).run(token.id, sourceId, card.excerpt || "", card.page_number || null);
+      ).run(
+        token.id,
+        cardSourceId,
+        card.excerpt || "",
+        card.page_number || null
+      );
     }
   });
   return { createdCount, linkedCount };
@@ -6069,6 +6146,14 @@ var BLOOM_VERBS2 = {
   4: "Analyze",
   5: "Synthesize"
 };
+var LlmResponseTruncatedError = class extends Error {
+  constructor(label) {
+    super(
+      `${label}: the model's response was cut off before producing usable output (finish_reason=length). The prompt likely left no room in the model's context window for a reply.`
+    );
+    this.name = "LlmResponseTruncatedError";
+  }
+};
 async function readChatContent(res, label) {
   if (!res.ok) {
     const errorText = await res.text().catch(() => "");
@@ -6077,11 +6162,16 @@ async function readChatContent(res, label) {
     );
   }
   const data = await res.json();
-  const content = data.choices?.[0]?.message?.content;
-  if (!content) {
+  const choice = data.choices?.[0];
+  const content = choice?.message?.content;
+  const trimmed = content?.trim() ?? "";
+  if (choice?.finish_reason === "length") {
+    throw new LlmResponseTruncatedError(label);
+  }
+  if (!trimmed) {
     throw new Error("Empty response from LLM");
   }
-  return content.trim();
+  return trimmed;
 }
 async function generateQuestionViaLLM(db, input8) {
   const cfg = await getProviderForRole(db, "recall");
@@ -6248,15 +6338,42 @@ function parseGeneratedCardArray(responseText, label, limits) {
     };
   });
 }
-async function importCurriculumViaLLM(db, text, targetCategory, sourceUrl) {
-  if (text.length > MAX_IMPORT_TEXT_CHARS) {
-    throw new Error(
-      `Curriculum text exceeds the ${MAX_IMPORT_TEXT_CHARS.toLocaleString()} character limit`
-    );
+var CURRICULUM_CHUNK_TEXT_CHARS = 3e3;
+function splitCurriculumText(text, maxChars) {
+  if (text.length <= maxChars) return [text];
+  const chunks = [];
+  let start = 0;
+  while (start < text.length) {
+    let end = Math.min(start + maxChars, text.length);
+    if (end < text.length) {
+      const minimumBoundary = start + Math.floor(maxChars * 0.6);
+      const sentenceWindow = text.slice(minimumBoundary, end);
+      let sentenceBoundary = -1;
+      for (const match of sentenceWindow.matchAll(/[.!?;:]\s+/g)) {
+        sentenceBoundary = (match.index ?? 0) + match[0].length;
+      }
+      if (sentenceBoundary > 0) {
+        end = minimumBoundary + sentenceBoundary;
+      } else {
+        const whitespaceBoundary = text.lastIndexOf(" ", end);
+        if (whitespaceBoundary >= minimumBoundary) end = whitespaceBoundary + 1;
+      }
+    }
+    chunks.push(text.slice(start, end));
+    start = end;
   }
-  const cfg = await getProviderForRole(db, "text");
-  const endpoint = await resolveUsableTextEndpoint(db);
-  const langName = LANGUAGE_NAMES[cfg.locale] || "English";
+  return chunks;
+}
+function deduplicateGeneratedCards(cards) {
+  const seen = /* @__PURE__ */ new Set();
+  return cards.filter((card) => {
+    const key = `${card.question.trim().toLocaleLowerCase()}\0${card.concept.trim().toLocaleLowerCase()}`;
+    if (seen.has(key)) return false;
+    seen.add(key);
+    return true;
+  });
+}
+async function requestCurriculumCards(endpoint, locale, langName, curriculumText, targetCategory, sourceUrl) {
   const systemPrompt = `You are ZAM, a highly precise agentic curriculum parser.
 Your task is to analyze curriculum objectives, syllabus requirements, or textbook notes, and extract atomic facts, concepts, or skills as structured learning cards in ${langName}.
 
@@ -6285,7 +6402,7 @@ Guidelines:
   curriculum section.
 - Output ONLY a raw valid JSON array of objects. Do NOT wrap the JSON in markdown code blocks, HTML, or include any preamble, headers, or conversational filler.`;
   const userPrompt = `Curriculum Text to Parse:
-${text}
+${curriculumText}
 
 Target Category: ${targetCategory}
 ${sourceUrl ? `Source Reference Link: ${sourceUrl}` : ""}
@@ -6308,10 +6425,73 @@ JSON Array Output:`;
         temperature: 0.1,
         max_tokens: DEFAULT_LLM_MAX_TOKENS
       }),
-      locale: cfg.locale
+      locale
     }
   );
-  const responseText = await readChatContent(res, "LLM curriculum import");
+  return readChatContent(res, "LLM curriculum import");
+}
+async function importCurriculumViaLLM(db, text, targetCategory, sourceUrl) {
+  if (text.length > MAX_IMPORT_TEXT_CHARS) {
+    throw new Error(
+      `Curriculum text exceeds the ${MAX_IMPORT_TEXT_CHARS.toLocaleString()} character limit`
+    );
+  }
+  const cfg = await getProviderForRole(db, "text");
+  const endpoint = await resolveUsableTextEndpoint(db);
+  const langName = LANGUAGE_NAMES[cfg.locale] || "English";
+  let responseText;
+  try {
+    responseText = await requestCurriculumCards(
+      endpoint,
+      cfg.locale,
+      langName,
+      text,
+      targetCategory,
+      sourceUrl
+    );
+  } catch (err) {
+    if (!(err instanceof LlmResponseTruncatedError) || text.length <= CURRICULUM_CHUNK_TEXT_CHARS) {
+      throw err;
+    }
+    const chunks = splitCurriculumText(text, CURRICULUM_CHUNK_TEXT_CHARS);
+    const cards = [];
+    for (const [index, chunk] of chunks.entries()) {
+      let chunkResponse;
+      try {
+        chunkResponse = await requestCurriculumCards(
+          endpoint,
+          cfg.locale,
+          langName,
+          chunk,
+          targetCategory,
+          sourceUrl
+        );
+      } catch (chunkError) {
+        if (chunkError instanceof LlmResponseTruncatedError) {
+          throw new Error(
+            `Curriculum chunk ${index + 1} of ${chunks.length} is still too large for the configured model's context window. Configure a model with a larger context window, or import a smaller source.`
+          );
+        }
+        throw chunkError;
+      }
+      cards.push(
+        ...parseGeneratedCardArray(chunkResponse, "curriculum import", {
+          min: 0,
+          max: 200
+        })
+      );
+    }
+    const deduplicated = deduplicateGeneratedCards(cards);
+    if (deduplicated.length > 200) {
+      throw new Error(
+        `Invalid curriculum import response: expected 0-200 cards, got ${deduplicated.length}`
+      );
+    }
+    return deduplicated.map((card) => ({
+      ...card,
+      source_link: sourceUrl || null
+    }));
+  }
   return parseGeneratedCardArray(responseText, "curriculum import", {
     min: 0,
     max: 200
@@ -7269,6 +7449,412 @@ async function readImageOCR(db, imagePath) {
   return extractTextFromScanViaLLM(db, imagePath);
 }
 
+// src/cli/curriculum/breadcrumb.ts
+var LAST_SELECTION_KEY = "curriculum.lastSelection";
+async function getLastCurriculumSelection(db) {
+  const raw = await getSetting(db, LAST_SELECTION_KEY);
+  if (!raw) return void 0;
+  try {
+    return JSON.parse(raw);
+  } catch {
+    return void 0;
+  }
+}
+async function setLastCurriculumSelection(db, breadcrumb) {
+  await setSetting(db, LAST_SELECTION_KEY, JSON.stringify(breadcrumb));
+}
+
+// src/cli/curriculum/providers/bildungsplan-bw/manifest.ts
+var BILDUNGSPLAN_BW_MANIFEST = {
+  schoolYear: "2026/2027",
+  capturedOn: "2026-07-02",
+  sourceRevision: "Bildungsplan Baden-W\xFCrttemberg Gymnasium 2016",
+  schoolTypes: [
+    { id: "gymnasium", label: "Gymnasium" },
+    { id: "realschule", label: "Realschule" }
+  ],
+  grades: {
+    gymnasium: ["9", "10", "11", "12"]
+  },
+  subjects: {
+    gymnasium: [
+      { id: "mathematik", label: "Mathematik" },
+      { id: "physik", label: "Physik" },
+      { id: "englisch", label: "Englisch" }
+    ]
+  },
+  tracks: {},
+  topics: {
+    "gymnasium|10|mathematik": [
+      { id: "leitidee-zahl", label: "Leitidee Zahl - Variable - Operation" },
+      { id: "leitidee-messung", label: "Leitidee Messen" },
+      { id: "leitidee-raum", label: "Leitidee Raum und Form" },
+      { id: "leitidee-funktion", label: "Leitidee Funktionaler Zusammenhang" },
+      { id: "leitidee-daten", label: "Leitidee Daten und Zufall" }
+    ]
+  },
+  contentUrls: {
+    "gymnasium|10|mathematik": "http://www.bildungsplaene-bw.de/,Lde/LS/BP2016BW/ALLG/GYM/M/bp/klasse-10"
+  }
+};
+
+// src/cli/curriculum/providers/bildungsplan-bw/index.ts
+var bildungsplanBwProvider = {
+  id: "bildungsplan-bw",
+  country: "DE",
+  countryLabel: "Deutschland",
+  region: "BW",
+  regionLabel: "Baden-W\xFCrttemberg",
+  label: "Bildungsplan (Baden-W\xFCrttemberg)",
+  listSchoolTypes() {
+    return BILDUNGSPLAN_BW_MANIFEST.schoolTypes;
+  },
+  listGrades(schoolType) {
+    return (BILDUNGSPLAN_BW_MANIFEST.grades[schoolType] || []).map((id) => ({
+      id,
+      label: `Klasse ${id}`
+    }));
+  },
+  listSubjects(schoolType, _grade) {
+    return BILDUNGSPLAN_BW_MANIFEST.subjects[schoolType] || [];
+  },
+  listTracks(schoolType, grade, subject) {
+    const key = `${schoolType}|${grade}|${subject}`;
+    return BILDUNGSPLAN_BW_MANIFEST.tracks[key] || [];
+  },
+  listTopics(selection) {
+    const key = selection.track ? `${selection.schoolType}|${selection.grade}|${selection.subject}|${selection.track}` : `${selection.schoolType}|${selection.grade}|${selection.subject}`;
+    const list = BILDUNGSPLAN_BW_MANIFEST.topics[key] || [];
+    return list.map((t2) => ({
+      ...t2,
+      sourceRef: key
+    }));
+  },
+  resolveTopic(topic) {
+    const uri = BILDUNGSPLAN_BW_MANIFEST.contentUrls[topic.sourceRef];
+    if (!uri) {
+      throw new Error(
+        `Bildungsplan Baden-W\xFCrttemberg: no resolvable source URL for topic "${topic.id}" (${topic.sourceRef}).`
+      );
+    }
+    return {
+      provider: "bildungsplan-bw",
+      topicId: `${topic.sourceRef}#${topic.id}`,
+      uri
+    };
+  },
+  extractTopics(html, topicIds) {
+    const chunks = html.split('<div class="bp-topic-block" id="bp_topic_');
+    const sections = [];
+    for (let i = 1; i < chunks.length; i++) {
+      const chunk = chunks[i];
+      const quoteIdx = chunk.indexOf('"');
+      if (quoteIdx === -1) continue;
+      const id = chunk.slice(0, quoteIdx);
+      const contentHtml = chunk.slice(quoteIdx + 1);
+      const headerMatch = contentHtml.match(
+        /<h[1-6][^>]*>([\s\S]*?)<\/h[1-6]>/i
+      );
+      const headerText = headerMatch ? cleanHtmlText(headerMatch[1]).trim() : "";
+      sections.push({
+        id,
+        headerText,
+        contentHtml
+      });
+    }
+    const results = {};
+    for (const topicId of topicIds) {
+      const hashIdx = topicId.indexOf("#");
+      if (hashIdx === -1) continue;
+      const key = topicId.substring(0, hashIdx);
+      const shortId = topicId.substring(hashIdx + 1);
+      const list = BILDUNGSPLAN_BW_MANIFEST.topics[key] ?? [];
+      const match = list.find((t2) => t2.id === shortId);
+      if (!match) continue;
+      const label = match.label;
+      const normalizedLabel = normalizeForComparison(label);
+      const sectionIndex = sections.findIndex((s) => {
+        const normalizedHeader = normalizeForComparison(s.headerText);
+        return normalizedHeader.includes(normalizedLabel) || s.id === shortId;
+      });
+      if (sectionIndex === -1) {
+        continue;
+      }
+      results[topicId] = cleanHtmlText(sections[sectionIndex].contentHtml);
+    }
+    return results;
+  }
+};
+function cleanHtmlText(html) {
+  let text = html.replace(
+    /<(head|script|style|svg)[^>]*>[\s\S]*?<\/\1>/gi,
+    " "
+  );
+  text = text.replace(/<h[1-6][^>]*>([\s\S]*?)<\/h[1-6]>/gi, "\n\n$1\n\n");
+  text = text.replace(/<li[^>]*>/gi, "\n- ");
+  text = text.replace(/<p[^>]*>/gi, "\n");
+  text = text.replace(/<br\s*\/?>/gi, "\n");
+  text = text.replace(/<[^>]+>/g, " ");
+  text = text.replace(/&nbsp;/g, " ").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&amp;/g, "&").replace(/&quot;/g, '"').replace(/&#39;/g, "'").replace(/&ndash;/g, "\u2013").replace(/&mdash;/g, "\u2014").replace(/&middot;/g, "\xB7");
+  return text.split("\n").map((line) => line.trim()).filter((line) => line.length > 0).join("\n");
+}
+function normalizeForComparison(str) {
+  return str.toLowerCase().replace(/&[a-z0-9#]+;/gi, "").replace(/[^a-z0-9]/gi, "");
+}
+
+// src/cli/curriculum/providers/lehrplanplus-bayern/manifest.ts
+var LEHRPLANPLUS_BAYERN_MANIFEST = {
+  schoolYear: "2026/2027",
+  capturedOn: "2026-07-02",
+  sourceRevision: "LehrplanPLUS Realschule \u2013 Oktober 2023",
+  schoolTypes: [
+    { id: "grundschule", label: "Grundschule" },
+    { id: "mittelschule", label: "Mittelschule" },
+    { id: "foerderschule", label: "F\xF6rderschule" },
+    { id: "realschule", label: "Realschule" },
+    { id: "gymnasium", label: "Gymnasium" },
+    { id: "wirtschaftsschule", label: "Wirtschaftsschule" },
+    { id: "fos", label: "Fachoberschule" },
+    { id: "bos", label: "Berufsoberschule" }
+  ],
+  grades: {
+    realschule: ["5", "6", "7", "8", "9", "10"]
+  },
+  subjects: {
+    realschule: [
+      {
+        id: "bwl-rechnungswesen",
+        label: "Betriebswirtschaftslehre / Rechnungswesen"
+      },
+      { id: "biologie", label: "Biologie" },
+      { id: "chemie", label: "Chemie" },
+      { id: "deutsch", label: "Deutsch" },
+      { id: "englisch", label: "Englisch" },
+      { id: "ernaehrung_und_gesundheit", label: "Ern\xE4hrung und Gesundheit" },
+      { id: "ethik", label: "Ethik" },
+      {
+        id: "evangelische-religionslehre",
+        label: "Evangelische Religionslehre"
+      },
+      { id: "franzoesisch", label: "Franz\xF6sisch" },
+      { id: "geographie", label: "Geographie" },
+      { id: "geschichte", label: "Geschichte" },
+      { id: "it", label: "Informationstechnologie" },
+      { id: "iu", label: "Islamischer Unterricht" },
+      { id: "ir", label: "Israelitische Religionslehre" },
+      {
+        id: "katholische-religionslehre",
+        label: "Katholische Religionslehre"
+      },
+      { id: "kunst", label: "Kunst" },
+      { id: "mathematik", label: "Mathematik" },
+      { id: "musik", label: "Musik" },
+      { id: "or", label: "Orthodoxe Religionslehre" },
+      { id: "physik", label: "Physik" },
+      { id: "pug", label: "Politik und Gesellschaft" },
+      { id: "soziallehre", label: "Soziallehre" },
+      { id: "sozialwesen", label: "Sozialwesen" },
+      { id: "spanisch", label: "Spanisch" },
+      { id: "sport", label: "Sport" },
+      { id: "textiles-gestalten", label: "Textiles Gestalten" },
+      { id: "werken", label: "Werken" },
+      { id: "wirtschaft-und-recht", label: "Wirtschaft und Recht" }
+    ]
+  },
+  tracks: {
+    "realschule|9|mathematik": [
+      { id: "wpfg1", label: "Mathematik 9 (I)" },
+      { id: "wpfg2-3", label: "Mathematik 9 (II/III)" }
+    ]
+  },
+  topics: {
+    "realschule|9|mathematik|wpfg1": [
+      { id: "lb1", label: "Reelle Zahlen", hours: 10 },
+      { id: "lb2", label: "Zentrische Streckung", hours: 17 },
+      { id: "lb3", label: "Rechtwinklige Dreiecke", hours: 20 },
+      { id: "lb4", label: "Kreis", hours: 10 },
+      { id: "lb5", label: "Raumgeometrie", hours: 20 },
+      { id: "lb6", label: "Systeme linearer Gleichungen", hours: 12 },
+      {
+        id: "lb7",
+        label: "Quadratische Funktionen und quadratische Gleichungen",
+        hours: 42
+      },
+      { id: "lb8", label: "Daten und Zufall", hours: 9 }
+    ],
+    "realschule|9|mathematik|wpfg2-3": [
+      { id: "lb1", label: "Reelle Zahlen", hours: 7 },
+      { id: "lb2", label: "Zentrische Streckung", hours: 13 },
+      { id: "lb3", label: "Rechtwinklige Dreiecke", hours: 20 },
+      { id: "lb4", label: "Kreis", hours: 10 },
+      { id: "lb5", label: "Lineare Funktionen", hours: 15 },
+      { id: "lb6", label: "Systeme linearer Gleichungen", hours: 10 },
+      { id: "lb7", label: "Daten und Zufall", hours: 9 }
+    ],
+    "realschule|9|deutsch": [
+      { id: "lb1", label: "Sprechen und Zuh\xF6ren" },
+      { id: "lb2", label: "Lesen \u2013 mit Texten und weiteren Medien umgehen" },
+      { id: "lb3", label: "Schreiben" },
+      {
+        id: "lb4",
+        label: "Sprachgebrauch und Sprache untersuchen und reflektieren"
+      }
+    ],
+    "realschule|9|englisch": [
+      { id: "lb1", label: "Kommunikative Kompetenzen" },
+      { id: "lb2", label: "Interkulturelle Kompetenzen" },
+      { id: "lb3", label: "Text- und Medienkompetenzen" },
+      { id: "lb4", label: "Methodische Kompetenzen" },
+      { id: "lb5", label: "Themengebiete" }
+    ]
+  },
+  contentUrls: {
+    "realschule|9|mathematik|wpfg1": "https://www.lehrplanplus.bayern.de/schulart/realschule/jgs/9/fach/mathematik/inhalt/fachlehrplaene?w_schulart=realschule&wt_1=schulart&w_fach=mathematik&wt_2=fach&w_jgs=9&wt_3=jgs&w_auspraegung=wpfg1",
+    "realschule|9|mathematik|wpfg2-3": "https://www.lehrplanplus.bayern.de/schulart/realschule/jgs/9/fach/mathematik/inhalt/fachlehrplaene?w_schulart=realschule&wt_1=schulart&w_fach=mathematik&wt_2=fach&w_jgs=9&wt_3=jgs&w_auspraegung=wpfg2-3",
+    "realschule|9|deutsch": "https://www.lehrplanplus.bayern.de/schulart/realschule/jgs/9/fach/deutsch/inhalt/fachlehrplaene",
+    "realschule|9|englisch": "https://www.lehrplanplus.bayern.de/schulart/realschule/jgs/9/fach/englisch/inhalt/fachlehrplaene"
+  }
+};
+
+// src/cli/curriculum/providers/lehrplanplus-bayern/index.ts
+function levelKey(schoolType, grade, subject, track) {
+  return track ? `${schoolType}|${grade}|${subject}|${track}` : `${schoolType}|${grade}|${subject}`;
+}
+var lehrplanplusBayernProvider = {
+  id: "lehrplanplus-bayern",
+  country: "DE",
+  countryLabel: "Deutschland",
+  region: "BY",
+  regionLabel: "Bayern",
+  label: "LehrplanPLUS (Bayern)",
+  listSchoolTypes() {
+    return LEHRPLANPLUS_BAYERN_MANIFEST.schoolTypes;
+  },
+  listGrades(schoolType) {
+    return (LEHRPLANPLUS_BAYERN_MANIFEST.grades[schoolType] ?? []).map((grade) => ({
+      id: grade,
+      label: grade
+    }));
+  },
+  listSubjects(schoolType, _grade) {
+    return LEHRPLANPLUS_BAYERN_MANIFEST.subjects[schoolType] ?? [];
+  },
+  listTracks(schoolType, grade, subject) {
+    return LEHRPLANPLUS_BAYERN_MANIFEST.tracks[levelKey(schoolType, grade, subject)] ?? [];
+  },
+  listTopics(selection) {
+    const { schoolType, grade, subject, track } = selection;
+    if (!schoolType || !grade || !subject) return [];
+    const key = levelKey(schoolType, grade, subject, track);
+    return (LEHRPLANPLUS_BAYERN_MANIFEST.topics[key] ?? []).map((topic) => ({
+      ...topic,
+      sourceRef: key
+    }));
+  },
+  resolveTopic(topic) {
+    const uri = LEHRPLANPLUS_BAYERN_MANIFEST.contentUrls[topic.sourceRef];
+    if (!uri) {
+      throw new Error(
+        `LehrplanPLUS Bayern: no resolvable source URL for topic "${topic.id}" (${topic.sourceRef}). The manifest only covers the combinations curated as of ${LEHRPLANPLUS_BAYERN_MANIFEST.capturedOn}.`
+      );
+    }
+    return {
+      provider: "lehrplanplus-bayern",
+      topicId: `${topic.sourceRef}#${topic.id}`,
+      uri
+    };
+  },
+  extractTopics(html, topicIds) {
+    const chunks = html.split('<div id="thema_');
+    const sections = [];
+    for (let i = 1; i < chunks.length; i++) {
+      const chunk = chunks[i];
+      const quoteIdx = chunk.indexOf('"');
+      if (quoteIdx === -1) continue;
+      const id = chunk.slice(0, quoteIdx);
+      const classStart = chunk.indexOf('class="');
+      if (classStart === -1) continue;
+      const classEnd = chunk.indexOf('"', classStart + 7);
+      const classContent = chunk.slice(classStart + 7, classEnd);
+      const lvlMatch = classContent.match(/headline_lvl(\d+)/);
+      if (!lvlMatch) continue;
+      const level = parseInt(lvlMatch[1], 10);
+      const contentHtml = chunk.slice(classEnd + 1);
+      const headerMatch = contentHtml.match(
+        /<a[^>]*class="paragraph_toggle"[^>]*>([\s\S]*?)<\/a>/i
+      );
+      const headerText = headerMatch ? cleanHtmlText2(headerMatch[1]).trim() : "";
+      sections.push({
+        id,
+        level,
+        headerText,
+        contentHtml
+      });
+    }
+    const results = {};
+    for (const topicId of topicIds) {
+      let label = "";
+      for (const key of Object.keys(LEHRPLANPLUS_BAYERN_MANIFEST.topics)) {
+        const list = LEHRPLANPLUS_BAYERN_MANIFEST.topics[key];
+        const match = list.find(
+          (t2) => t2.id === topicId || `${key}#${t2.id}` === topicId
+        );
+        if (match) {
+          label = match.label;
+          break;
+        }
+      }
+      if (!label) {
+        continue;
+      }
+      const normalizedLabel = normalizeForComparison2(label);
+      const lvl1Index = sections.findIndex((s) => {
+        if (s.level !== 1) return false;
+        const normalizedHeader = normalizeForComparison2(s.headerText);
+        return normalizedHeader.includes(normalizedLabel);
+      });
+      if (lvl1Index === -1) {
+        continue;
+      }
+      const collectedSections = [sections[lvl1Index]];
+      for (let i = lvl1Index + 1; i < sections.length; i++) {
+        if (sections[i].level === 1) {
+          break;
+        }
+        collectedSections.push(sections[i]);
+      }
+      const fullHtml = collectedSections.map((s) => s.contentHtml).join("\n");
+      results[topicId] = cleanHtmlText2(fullHtml);
+    }
+    return results;
+  }
+};
+function cleanHtmlText2(html) {
+  let text = html.replace(
+    /<(head|script|style|svg)[^>]*>[\s\S]*?<\/\1>/gi,
+    " "
+  );
+  text = text.replace(/<h[1-6][^>]*>([\s\S]*?)<\/h[1-6]>/gi, "\n\n$1\n\n");
+  text = text.replace(/<li[^>]*>/gi, "\n- ");
+  text = text.replace(/<p[^>]*>/gi, "\n");
+  text = text.replace(/<br\s*\/?>/gi, "\n");
+  text = text.replace(/<[^>]+>/g, " ");
+  text = text.replace(/&nbsp;/g, " ").replace(/&lt;/g, "<").replace(/&gt;/g, ">").replace(/&amp;/g, "&").replace(/&quot;/g, '"').replace(/&#39;/g, "'").replace(/&ndash;/g, "\u2013").replace(/&mdash;/g, "\u2014").replace(/&middot;/g, "\xB7");
+  return text.split("\n").map((line) => line.trim()).filter((line) => line.length > 0).join("\n");
+}
+function normalizeForComparison2(str) {
+  return str.toLowerCase().replace(/&[a-z0-9#]+;/gi, "").replace(/[^a-z0-9]/gi, "");
+}
+
+// src/cli/curriculum/registry.ts
+var CURRICULUM_PROVIDERS = [
+  lehrplanplusBayernProvider,
+  bildungsplanBwProvider
+];
+function getCurriculumProvider(id) {
+  return CURRICULUM_PROVIDERS.find((provider) => provider.id === id);
+}
+
 // src/cli/llm/vision.ts
 import { randomBytes } from "crypto";
 import { readFileSync as readFileSync10 } from "fs";
@@ -10555,8 +11141,22 @@ bridgeCommand.command("personal-card-confirm-foundations").description("Save con
 });
 bridgeCommand.command("personal-source-import").description(
   "Fetch and clean plain text from local file, web link, or vision scan (JSON)"
-).requiredOption("--type <file|web|scan>", "Source type").requiredOption("--uri <uri>", "Source path or URL").action(async (opts) => {
+).requiredOption("--type <file|web|scan>", "Source type").requiredOption("--uri <uri>", "Source path or URL").option("--refresh", "Re-fetch an already cached web source").action(async (opts) => {
   await withDb2(async (db) => {
+    if (opts.type === "web" && !opts.refresh) {
+      const cached = await db.prepare(
+        "SELECT id, content FROM sources WHERE uri = ? AND type = 'web'"
+      ).get(opts.uri);
+      if (cached?.content) {
+        jsonOut2({
+          success: true,
+          sourceId: cached.id,
+          content: cached.content,
+          cached: true
+        });
+        return;
+      }
+    }
     let content = "";
     if (opts.type === "file") {
       content = await readLocalFile(opts.uri);
@@ -10579,7 +11179,8 @@ bridgeCommand.command("personal-source-import").description(
     jsonOut2({
       success: true,
       sourceId: record.id,
-      content: record.content
+      content: record.content,
+      cached: false
     });
   });
 });
@@ -10605,6 +11206,205 @@ bridgeCommand.command("personal-source-confirm-import").description(
     });
   });
 });
+bridgeCommand.command("curriculum-list-providers").description("List registered curriculum providers (JSON)").action(() => {
+  jsonOut2({
+    success: true,
+    providers: CURRICULUM_PROVIDERS.map((provider) => ({
+      id: provider.id,
+      country: provider.country,
+      countryLabel: provider.countryLabel,
+      region: provider.region,
+      regionLabel: provider.regionLabel,
+      label: provider.label
+    }))
+  });
+});
+var CURRICULUM_LEVELS = [
+  "schoolType",
+  "grade",
+  "subject",
+  "track",
+  "topic"
+];
+bridgeCommand.command("curriculum-list-level").description("List taxonomy options for the next import-wizard step (JSON)").requiredOption("--provider <id>", "Curriculum provider id").requiredOption(
+  "--level <level>",
+  `Level to list: ${CURRICULUM_LEVELS.join("|")}`
+).option("--selection <json>", "JSON selection made so far", "{}").action((opts) => {
+  const provider = getCurriculumProvider(opts.provider);
+  if (!provider) jsonError(`Unknown curriculum provider: ${opts.provider}`);
+  if (!CURRICULUM_LEVELS.includes(opts.level)) {
+    jsonError(
+      `Invalid --level: ${opts.level}. Use one of ${CURRICULUM_LEVELS.join(", ")}.`
+    );
+  }
+  let selection;
+  try {
+    selection = JSON.parse(opts.selection);
+  } catch {
+    jsonError("Invalid --selection JSON");
+    return;
+  }
+  const level = opts.level;
+  if (level === "schoolType") {
+    jsonOut2({ success: true, options: provider.listSchoolTypes() });
+    return;
+  }
+  if (!selection.schoolType) jsonError("selection.schoolType is required");
+  if (level === "grade") {
+    jsonOut2({
+      success: true,
+      options: provider.listGrades(selection.schoolType)
+    });
+    return;
+  }
+  if (!selection.grade) jsonError("selection.grade is required");
+  if (level === "subject") {
+    jsonOut2({
+      success: true,
+      options: provider.listSubjects(selection.schoolType, selection.grade)
+    });
+    return;
+  }
+  if (!selection.subject) jsonError("selection.subject is required");
+  if (level === "track") {
+    jsonOut2({
+      success: true,
+      options: provider.listTracks(
+        selection.schoolType,
+        selection.grade,
+        selection.subject
+      )
+    });
+    return;
+  }
+  jsonOut2({ success: true, options: provider.listTopics(selection) });
+});
+bridgeCommand.command("curriculum-resolve-topics").description("Resolve selected curriculum topics to source URLs (JSON)").requiredOption("--provider <id>", "Curriculum provider id").requiredOption("--topics <json>", "JSON array of topic nodes to resolve").action((opts) => {
+  const provider = getCurriculumProvider(opts.provider);
+  if (!provider) jsonError(`Unknown curriculum provider: ${opts.provider}`);
+  let topics;
+  try {
+    topics = JSON.parse(opts.topics);
+  } catch {
+    jsonError("Invalid --topics JSON");
+    return;
+  }
+  const resolved = topics.map((topic) => provider.resolveTopic(topic));
+  jsonOut2({ success: true, resolved });
+});
+async function fetchRawHtml(url) {
+  if (!await isSafeUrl(url)) {
+    throw new Error(`Access denied to unsafe target URL: ${url}`);
+  }
+  const controller = new AbortController();
+  const timeoutId = setTimeout(() => controller.abort(), 1e4);
+  try {
+    const res = await fetch(url, {
+      signal: controller.signal,
+      headers: {
+        "User-Agent": "ZAM-Content-Studio/0.6.1"
+      }
+    });
+    if (!res.ok) {
+      throw new Error(`Web server responded with status ${res.status}`);
+    }
+    const contentType = res.headers.get("content-type") || "";
+    if (!contentType.includes("text/html") && !contentType.includes("application/xhtml+xml")) {
+      throw new Error(`Unsupported content type: ${contentType}`);
+    }
+    return await res.text();
+  } catch (err) {
+    if (err instanceof Error && err.name === "AbortError") {
+      throw new Error("Connection request timed out after 10 seconds");
+    }
+    throw err;
+  } finally {
+    clearTimeout(timeoutId);
+  }
+}
+bridgeCommand.command("curriculum-extract-topics").description(
+  "Fetch and extract specific texts for selected curriculum topics (JSON)"
+).requiredOption("--provider <id>", "Curriculum provider id").requiredOption("--topics <json>", "JSON array of selected topic nodes").action(async (opts) => {
+  await withDb2(async (db) => {
+    const provider = getCurriculumProvider(opts.provider);
+    if (!provider) jsonError(`Unknown curriculum provider: ${opts.provider}`);
+    let topics;
+    try {
+      topics = JSON.parse(opts.topics);
+    } catch {
+      jsonError("Invalid --topics JSON");
+      return;
+    }
+    const topicsByUri = /* @__PURE__ */ new Map();
+    for (const topic of topics) {
+      const resolved = provider.resolveTopic(topic);
+      const list = topicsByUri.get(resolved.uri) || [];
+      list.push(topic);
+      topicsByUri.set(resolved.uri, list);
+    }
+    const extracted = [];
+    for (const [uri, uriTopics] of topicsByUri.entries()) {
+      const rawHtml = await fetchRawHtml(uri);
+      let extractedTexts = {};
+      if (provider.extractTopics) {
+        const fullTopicIds = uriTopics.map((t2) => `${t2.sourceRef}#${t2.id}`);
+        extractedTexts = provider.extractTopics(rawHtml, fullTopicIds);
+      } else {
+        const cleanText = cleanHtml(rawHtml);
+        for (const t2 of uriTopics) {
+          extractedTexts[`${t2.sourceRef}#${t2.id}`] = cleanText;
+        }
+      }
+      const pageCleanedText = cleanHtml(rawHtml);
+      const sourceId = ulid8();
+      await db.prepare(
+        `INSERT INTO sources (id, type, uri, content)
+             VALUES (?, 'web', ?, ?)
+             ON CONFLICT(uri) DO UPDATE SET
+               type = excluded.type,
+               content = excluded.content`
+      ).run(sourceId, uri, pageCleanedText);
+      const record = await db.prepare("SELECT id FROM sources WHERE uri = ?").get(uri);
+      for (const topic of uriTopics) {
+        const resolved = provider.resolveTopic(topic);
+        const text = extractedTexts[resolved.topicId] || "";
+        extracted.push({
+          topicId: resolved.topicId,
+          uri,
+          sourceId: record.id,
+          text
+        });
+      }
+    }
+    jsonOut2({ success: true, extracted });
+  });
+});
+bridgeCommand.command("curriculum-get-last-selection").description("Read the learner's last navigated curriculum path (JSON)").action(async () => {
+  await withDb2(async (db) => {
+    const breadcrumb = await getLastCurriculumSelection(db);
+    jsonOut2({ success: true, breadcrumb: breadcrumb ?? null });
+  });
+});
+bridgeCommand.command("curriculum-set-last-selection").description("Persist the learner's last navigated curriculum path (JSON)").requiredOption(
+  "--breadcrumb <json>",
+  "JSON breadcrumb: {providerId, schoolType?, grade?, subject?, track?}"
+).action(async (opts) => {
+  await withDb2(async (db) => {
+    let breadcrumb;
+    try {
+      breadcrumb = JSON.parse(opts.breadcrumb);
+    } catch {
+      jsonError("Invalid --breadcrumb JSON");
+      return;
+    }
+    if (!breadcrumb.providerId) {
+      jsonError("breadcrumb.providerId is required");
+      return;
+    }
+    await setLastCurriculumSelection(db, breadcrumb);
+    jsonOut2({ success: true });
+  });
+});
 bridgeCommand.command("serve").description("Start the persistent JSON-RPC stdin/stdout server").option("--stdin", "Use stdin/stdout for communication").action(async (_opts) => {
   isServeMode = true;
   const {