unrag 0.2.5 → 0.2.7

package/dist/cli/index.js

@@ -13,7 +13,7 @@ import {
   select,
   text
 } from "@clack/prompts";
-import path5 from "node:path";
+import path6 from "node:path";
 import { fileURLToPath } from "node:url";
 
 // cli/lib/registry.ts
@@ -108,10 +108,31 @@ var EXTRACTOR_FLAG_KEYS = {
   "file-xlsx": ["file_xlsx"]
 };
 var ALL_FLAG_KEYS = Array.from(new Set(Object.values(EXTRACTOR_FLAG_KEYS).flat())).sort();
+var indentBlock = (text, spaces) => {
+  const pad = " ".repeat(spaces);
+  return text.split(`
+`).map((l) => l ? pad + l : l).join(`
+`);
+};
+var replaceBetweenMarkers = (content, startMarker, endMarker, replacement) => {
+  const startIdx = content.indexOf(startMarker);
+  const endIdx = content.indexOf(endMarker);
+  if (startIdx < 0 || endIdx < 0 || endIdx < startIdx)
+    return content;
+  const startLineStart = content.lastIndexOf(`
+`, startIdx);
+  const start = startLineStart < 0 ? 0 : startLineStart + 1;
+  const endLineEnd = content.indexOf(`
+`, endIdx);
+  const end = endLineEnd < 0 ? content.length : endLineEnd + 1;
+  return content.slice(0, start) + replacement + content.slice(end);
+};
 var renderUnragConfig = (content, selection) => {
   const installImportBase = `./${selection.installDir.replace(/\\/g, "/")}`;
   const richMedia = selection.richMedia ?? { enabled: false, extractors: [] };
   const selectedExtractors = Array.from(new Set(richMedia.extractors ?? [])).sort();
+  const preset = selection.presetConfig;
+  const embeddingProvider = selection.embeddingProvider ?? (typeof preset?.embedding?.provider === "string" ? preset.embedding.provider : undefined) ?? "ai";
   const baseImports = [
     `import { defineUnragConfig } from "${installImportBase}/core";`
   ];
@@ -145,17 +166,109 @@ var renderUnragConfig = (content, selection) => {
   ].join(`
 `);
   let out = content.replace("// __UNRAG_IMPORTS__", importsBlock).replace("// __UNRAG_CREATE_ENGINE__", createEngineBlock);
-  out = out.replace('type: "text", // __UNRAG_EMBEDDING_TYPE__', richMedia.enabled ? 'type: "multimodal",' : 'type: "text",').replace('model: "openai/text-embedding-3-small", // __UNRAG_EMBEDDING_MODEL__', richMedia.enabled ? 'model: "cohere/embed-v4.0",' : 'model: "openai/text-embedding-3-small",');
-  const enabledFlagKeys = new Set;
-  if (richMedia.enabled) {
-    for (const ex of selectedExtractors) {
-      for (const k of EXTRACTOR_FLAG_KEYS[ex] ?? []) {
-        enabledFlagKeys.add(k);
-      }
+  const presetChunkSize = preset?.defaults?.chunking?.chunkSize;
+  const presetChunkOverlap = preset?.defaults?.chunking?.chunkOverlap;
+  const presetTopK = preset?.defaults?.retrieval?.topK;
+  if (typeof presetChunkSize === "number") {
+    out = out.replace("chunkSize: 200, // __UNRAG_DEFAULT_chunkSize__", `chunkSize: ${presetChunkSize},`);
+  } else {
+    out = out.replace("chunkSize: 200, // __UNRAG_DEFAULT_chunkSize__", "chunkSize: 200,");
+  }
+  if (typeof presetChunkOverlap === "number") {
+    out = out.replace("chunkOverlap: 40, // __UNRAG_DEFAULT_chunkOverlap__", `chunkOverlap: ${presetChunkOverlap},`);
+  } else {
+    out = out.replace("chunkOverlap: 40, // __UNRAG_DEFAULT_chunkOverlap__", "chunkOverlap: 40,");
+  }
+  if (typeof presetTopK === "number") {
+    out = out.replace("topK: 8, // __UNRAG_DEFAULT_topK__", `topK: ${presetTopK},`);
+  } else {
+    out = out.replace("topK: 8, // __UNRAG_DEFAULT_topK__", "topK: 8,");
+  }
+  const presetEmbeddingType = preset?.embedding?.config?.type;
+  const presetEmbeddingModel = preset?.embedding?.config?.model;
+  const presetEmbeddingTimeoutMs = preset?.embedding?.config?.timeoutMs;
+  const providerLine = `    provider: "${embeddingProvider}",`;
+  out = out.replace(/^\s*provider:\s*".*?",\s*$/m, providerLine);
+  const defaultModelByProvider = {
+    ai: "openai/text-embedding-3-small",
+    openai: "text-embedding-3-small",
+    google: "gemini-embedding-001",
+    openrouter: "text-embedding-3-small",
+    azure: "text-embedding-3-small",
+    vertex: "text-embedding-004",
+    bedrock: "amazon.titan-embed-text-v2:0",
+    cohere: "embed-english-v3.0",
+    mistral: "mistral-embed",
+    together: "togethercomputer/m2-bert-80M-2k-retrieval",
+    ollama: "nomic-embed-text",
+    voyage: "voyage-3.5-lite",
+    custom: "openai/text-embedding-3-small"
+  };
+  const resolvedEmbeddingModel = (() => {
+    if (typeof presetEmbeddingModel === "string" && presetEmbeddingModel.trim().length > 0) {
+      return presetEmbeddingModel.trim();
+    }
+    if (embeddingProvider === "ai" && presetEmbeddingType === "multimodal") {
+      return "cohere/embed-v4.0";
+    }
+    return defaultModelByProvider[embeddingProvider] ?? "openai/text-embedding-3-small";
+  })();
+  const normalizeModelForProvider = (model) => {
+    if (embeddingProvider === "ai")
+      return model;
+    const prefix = `${embeddingProvider}/`;
+    return model.startsWith(prefix) ? model.slice(prefix.length) : model;
+  };
+  const nextModel = normalizeModelForProvider(resolvedEmbeddingModel);
+  out = out.replace('model: "openai/text-embedding-3-small", // __UNRAG_EMBEDDING_MODEL__', `model: ${JSON.stringify(nextModel)},`);
+  if (presetEmbeddingType === "multimodal") {
+    if (!out.includes('type: "multimodal"') && !out.includes('type: "text"')) {
+      out = out.replace(`config: {
+      model:`, `config: {
+      type: "multimodal",
+      model:`);
+    } else {
+      out = out.replace(/^\s*type:\s*".*?",\s*$/m, `      type: "multimodal",`);
     }
   }
-  for (const k of ALL_FLAG_KEYS) {
-    out = out.replace(`enabled: false, // __UNRAG_FLAG_${k}__`, `enabled: ${enabledFlagKeys.has(k) ? "true" : "false"},`);
+  if (typeof presetEmbeddingTimeoutMs === "number") {
+    out = out.replace("timeoutMs: 15_000, // __UNRAG_EMBEDDING_TIMEOUT__", `timeoutMs: ${presetEmbeddingTimeoutMs},`);
+  } else {
+    out = out.replace("timeoutMs: 15_000, // __UNRAG_EMBEDDING_TIMEOUT__", "timeoutMs: 15_000,");
+  }
+  const presetStoreChunkContent = preset?.engine?.storage?.storeChunkContent;
+  const presetStoreDocumentContent = preset?.engine?.storage?.storeDocumentContent;
+  if (typeof presetStoreChunkContent === "boolean") {
+    out = out.replace("storeChunkContent: true, // __UNRAG_STORAGE_storeChunkContent__", `storeChunkContent: ${presetStoreChunkContent},`);
+  } else {
+    out = out.replace("storeChunkContent: true, // __UNRAG_STORAGE_storeChunkContent__", "storeChunkContent: true,");
+  }
+  if (typeof presetStoreDocumentContent === "boolean") {
+    out = out.replace("storeDocumentContent: true, // __UNRAG_STORAGE_storeDocumentContent__", `storeDocumentContent: ${presetStoreDocumentContent},`);
+  } else {
+    out = out.replace("storeDocumentContent: true, // __UNRAG_STORAGE_storeDocumentContent__", "storeDocumentContent: true,");
+  }
+  const assetProcessingOverride = preset?.engine?.assetProcessing;
+  if (assetProcessingOverride && typeof assetProcessingOverride === "object") {
+    const json = JSON.stringify(assetProcessingOverride, null, 2);
+    const block = `  assetProcessing: ${indentBlock(json, 2).trimStart()},
+`;
+    out = replaceBetweenMarkers(out, "__UNRAG_ASSET_PROCESSING_BLOCK_START__", "__UNRAG_ASSET_PROCESSING_BLOCK_END__", block);
+  } else {
+    out = out.replace("// __UNRAG_ASSET_PROCESSING_BLOCK_START__", "").replace("// __UNRAG_ASSET_PROCESSING_BLOCK_END__", "");
+  }
+  if (!(assetProcessingOverride && typeof assetProcessingOverride === "object")) {
+    const enabledFlagKeys = new Set;
+    if (richMedia.enabled) {
+      for (const ex of selectedExtractors) {
+        for (const k of EXTRACTOR_FLAG_KEYS[ex] ?? []) {
+          enabledFlagKeys.add(k);
+        }
+      }
+    }
+    for (const k of ALL_FLAG_KEYS) {
+      out = out.replace(`enabled: false, // __UNRAG_FLAG_${k}__`, `enabled: ${enabledFlagKeys.has(k) ? "true" : "false"},`);
+    }
   }
   const extractorLines = richMedia.enabled && selectedExtractors.length > 0 ? selectedExtractors.map((ex) => `      ${EXTRACTOR_FACTORY[ex]}(),`).join(`
 `) : "";
@@ -164,6 +277,7 @@ var renderUnragConfig = (content, selection) => {
 };
 var renderDocs = (content, selection) => {
   const notes = [];
+  const embeddingProvider = selection.embeddingProvider ?? "ai";
   if (selection.storeAdapter === "drizzle") {
     notes.push("## Store adapter: Drizzle", "", "You can import the generated Drizzle schema module into your app’s main Drizzle schema to avoid duplicating table definitions.", "", "Example pattern:", "```ts", `import * as rag from "./${selection.installDir}/store/drizzle/schema";`, "", "export const schema = {", "  ...rag.schema,", "  // ...your app tables", "};", "```", "", "Then run Drizzle migrations from your app as usual.");
   } else if (selection.storeAdapter === "prisma") {
@@ -171,7 +285,42 @@ var renderDocs = (content, selection) => {
   } else {
     notes.push("## Store adapter: Raw SQL", "", "This adapter uses a `pg` Pool and parameterized SQL queries against the tables described above.", "It’s the most portable option when you don’t want ORM coupling.");
   }
-  const withNotes = content.replace("<!-- __UNRAG_ADAPTER_NOTES__ -->", notes.join(`
+  const envLines = [
+    "## Environment variables",
+    "",
+    "Add these to your environment:",
+    "- `DATABASE_URL` (Postgres connection string)"
+  ];
+  if (embeddingProvider === "ai") {
+    envLines.push("- `AI_GATEWAY_API_KEY` (required by the AI SDK when using Vercel AI Gateway)", "- Optional: `AI_GATEWAY_MODEL` (defaults to `openai/text-embedding-3-small`)");
+  } else if (embeddingProvider === "openai") {
+    envLines.push("- `OPENAI_API_KEY`", "- Optional: `OPENAI_EMBEDDING_MODEL` (defaults to `text-embedding-3-small`)");
+  } else if (embeddingProvider === "google") {
+    envLines.push("- `GOOGLE_GENERATIVE_AI_API_KEY`", "- Optional: `GOOGLE_GENERATIVE_AI_EMBEDDING_MODEL` (defaults to `gemini-embedding-001`)");
+  } else if (embeddingProvider === "openrouter") {
+    envLines.push("- `OPENROUTER_API_KEY`", "- Optional: `OPENROUTER_EMBEDDING_MODEL` (defaults to `text-embedding-3-small`)");
+  } else if (embeddingProvider === "cohere") {
+    envLines.push("- `COHERE_API_KEY`", "- Optional: `COHERE_EMBEDDING_MODEL` (defaults to `embed-english-v3.0`)");
+  } else if (embeddingProvider === "mistral") {
+    envLines.push("- `MISTRAL_API_KEY`", "- Optional: `MISTRAL_EMBEDDING_MODEL` (defaults to `mistral-embed`)");
+  } else if (embeddingProvider === "together") {
+    envLines.push("- `TOGETHER_AI_API_KEY`", "- Optional: `TOGETHER_AI_EMBEDDING_MODEL` (defaults to `togethercomputer/m2-bert-80M-2k-retrieval`)");
+  } else if (embeddingProvider === "voyage") {
+    envLines.push("- `VOYAGE_API_KEY`", "- Optional: `VOYAGE_MODEL` (defaults to `voyage-3.5-lite`)");
+  } else if (embeddingProvider === "ollama") {
+    envLines.push("- Optional: `OLLAMA_EMBEDDING_MODEL` (defaults to `nomic-embed-text`)");
+  } else if (embeddingProvider === "azure") {
+    envLines.push("- `AZURE_OPENAI_API_KEY`", "- `AZURE_RESOURCE_NAME`", "- Optional: `AZURE_EMBEDDING_MODEL` (defaults to `text-embedding-3-small`)");
+  } else if (embeddingProvider === "vertex") {
+    envLines.push("- `GOOGLE_APPLICATION_CREDENTIALS` (when running outside GCP)", "- Optional: `GOOGLE_VERTEX_EMBEDDING_MODEL` (defaults to `text-embedding-004`)");
+  } else if (embeddingProvider === "bedrock") {
+    envLines.push("- `AWS_REGION`", "- AWS credentials (`AWS_ACCESS_KEY_ID` / `AWS_SECRET_ACCESS_KEY`) when running outside AWS", "- Optional: `BEDROCK_EMBEDDING_MODEL` (defaults to `amazon.titan-embed-text-v2:0`)");
+  }
+  const withEnv = content.replace(/## Environment variables[\s\S]*?## Database requirements/, `${envLines.join(`
+`)}
+
+## Database requirements`);
+  const withNotes = withEnv.replace("<!-- __UNRAG_ADAPTER_NOTES__ -->", notes.join(`
 `));
   return withNotes.replaceAll("@unrag/config", `${selection.aliasBase}/config`).replaceAll("`@unrag/*`", `\`${selection.aliasBase}/*\``);
 };
@@ -225,9 +374,57 @@ async function copyRegistryFiles(selection) {
       src: path2.join(selection.registryRoot, "core/retrieve.ts"),
       dest: path2.join(installBaseAbs, "core/retrieve.ts")
     },
+    {
+      src: path2.join(selection.registryRoot, "embedding/_shared.ts"),
+      dest: path2.join(installBaseAbs, "embedding/_shared.ts")
+    },
     {
       src: path2.join(selection.registryRoot, "embedding/ai.ts"),
       dest: path2.join(installBaseAbs, "embedding/ai.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/openai.ts"),
+      dest: path2.join(installBaseAbs, "embedding/openai.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/google.ts"),
+      dest: path2.join(installBaseAbs, "embedding/google.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/openrouter.ts"),
+      dest: path2.join(installBaseAbs, "embedding/openrouter.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/azure.ts"),
+      dest: path2.join(installBaseAbs, "embedding/azure.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/vertex.ts"),
+      dest: path2.join(installBaseAbs, "embedding/vertex.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/bedrock.ts"),
+      dest: path2.join(installBaseAbs, "embedding/bedrock.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/cohere.ts"),
+      dest: path2.join(installBaseAbs, "embedding/cohere.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/mistral.ts"),
+      dest: path2.join(installBaseAbs, "embedding/mistral.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/together.ts"),
+      dest: path2.join(installBaseAbs, "embedding/together.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/ollama.ts"),
+      dest: path2.join(installBaseAbs, "embedding/ollama.ts")
+    },
+    {
+      src: path2.join(selection.registryRoot, "embedding/voyage.ts"),
+      dest: path2.join(installBaseAbs, "embedding/voyage.ts")
     }
   ];
   if (selection.storeAdapter === "drizzle") {
@@ -258,21 +455,27 @@ async function copyRegistryFiles(selection) {
       dest: path2.join(installBaseAbs, "store/prisma/store.ts")
     });
   }
+  const nonInteractive = Boolean(selection.yes) || !process.stdin.isTTY;
+  const overwritePolicy = selection.overwrite ?? "skip";
   for (const mapping of fileMappings) {
     if (!await exists(mapping.src)) {
       throw new Error(`Registry file missing: ${mapping.src}`);
     }
     if (await exists(mapping.dest)) {
-      const answer = await confirm({
-        message: `Overwrite ${path2.relative(selection.projectRoot, mapping.dest)}?`,
-        initialValue: false
-      });
-      if (isCancel(answer)) {
-        cancel("Cancelled.");
-        return;
-      }
-      if (!answer) {
+      if (overwritePolicy === "force") {} else if (nonInteractive) {
         continue;
+      } else {
+        const answer = await confirm({
+          message: `Overwrite ${path2.relative(selection.projectRoot, mapping.dest)}?`,
+          initialValue: false
+        });
+        if (isCancel(answer)) {
+          cancel("Cancelled.");
+          return;
+        }
+        if (!answer) {
+          continue;
+        }
       }
     }
     const raw = await readText(mapping.src);
@@ -290,6 +493,7 @@ async function copyConnectorFiles(selection) {
   const files = await listFilesRecursive(connectorRegistryAbs);
   const destRootAbs = path2.join(installBaseAbs, "connectors", selection.connector);
   const nonInteractive = Boolean(selection.yes) || !process.stdin.isTTY;
+  const overwritePolicy = selection.overwrite ?? "skip";
   for (const src of files) {
     if (!await exists(src)) {
       throw new Error(`Registry file missing: ${src}`);
@@ -297,7 +501,7 @@ async function copyConnectorFiles(selection) {
     const rel = path2.relative(connectorRegistryAbs, src);
     const dest = path2.join(destRootAbs, rel);
     if (await exists(dest)) {
-      if (nonInteractive) {
+      if (overwritePolicy === "force") {} else if (nonInteractive) {
         continue;
       }
       const answer = await confirm({
@@ -329,9 +533,12 @@ async function copyExtractorFiles(selection) {
   const destRootAbs = path2.join(installBaseAbs, "extractors", selection.extractor);
   const sharedDestRootAbs = path2.join(installBaseAbs, "extractors", "_shared");
   const nonInteractive = Boolean(selection.yes) || !process.stdin.isTTY;
+  const overwritePolicy = selection.overwrite ?? "skip";
   const shouldWrite = async (src, dest) => {
     if (!await exists(dest))
       return true;
+    if (overwritePolicy === "force")
+      return true;
     if (nonInteractive)
       return false;
     try {
@@ -388,26 +595,103 @@ async function writeJsonFile(filePath, data) {
 `, "utf8");
 }
 
-// cli/lib/packageJson.ts
+// cli/lib/manifest.ts
 import path3 from "node:path";
-import { readFile as readFile3, writeFile as writeFile3 } from "node:fs/promises";
+import { readFile as readFile3 } from "node:fs/promises";
+async function readRegistryManifest(registryRoot) {
+  const abs = path3.join(registryRoot, "manifest.json");
+  const raw = await readFile3(abs, "utf8");
+  const parsed = JSON.parse(raw);
+  if (!parsed || parsed.version !== 1) {
+    throw new Error(`Unsupported registry manifest version in ${abs}`);
+  }
+  if (!Array.isArray(parsed.extractors) || !Array.isArray(parsed.connectors)) {
+    throw new Error(`Invalid registry manifest shape in ${abs}`);
+  }
+  return parsed;
+}
+
+// cli/lib/constants.ts
+var UNRAG_SITE_URL = (process.env.UNRAG_SITE_URL ?? process.env.UNRAG_DOCS_BASE_URL)?.trim() || "https://unrag.dev";
+var UNRAG_GITHUB_REPO_URL = "https://github.com/BetterStacks/unrag";
+function docsUrl(siteRelativePath) {
+  const p = siteRelativePath.startsWith("/") ? siteRelativePath : `/${siteRelativePath}`;
+  const base = UNRAG_SITE_URL.endsWith("/") ? UNRAG_SITE_URL : `${UNRAG_SITE_URL}/`;
+  return new URL(p.replace(/^\/+/, "/"), base).toString();
+}
+
+// cli/lib/preset.ts
+function isPresetPayloadV1(x) {
+  if (!x || typeof x !== "object")
+    return false;
+  const o = x;
+  if (o.version !== 1)
+    return false;
+  if (!o.install || typeof o.install !== "object")
+    return false;
+  if (!o.modules || typeof o.modules !== "object")
+    return false;
+  if (typeof o.install.installDir !== "string")
+    return false;
+  if (!["drizzle", "prisma", "raw-sql"].includes(String(o.install.storeAdapter)))
+    return false;
+  if (typeof o.install.aliasBase !== "string")
+    return false;
+  if (!Array.isArray(o.modules.extractors) || !Array.isArray(o.modules.connectors))
+    return false;
+  return true;
+}
+function toPresetUrl(input) {
+  const s = String(input ?? "").trim();
+  if (!s) {
+    throw new Error("Missing preset id/url");
+  }
+  if (s.startsWith("http://") || s.startsWith("https://"))
+    return s;
+  return docsUrl(`/api/presets/${encodeURIComponent(s)}`);
+}
+async function fetchPreset(input) {
+  const url = toPresetUrl(input);
+  if (typeof fetch !== "function") {
+    throw new Error(`Global fetch() is unavailable in this runtime; cannot fetch preset from ${url}. Set UNRAG_SITE_URL="${UNRAG_SITE_URL}" and use a newer Node runtime.`);
+  }
+  const res = await fetch(url, {
+    headers: {
+      "user-agent": "unrag-cli",
+      accept: "application/json"
+    }
+  });
+  if (!res.ok) {
+    const text = await res.text().catch(() => "");
+    throw new Error(`Failed to fetch preset (${res.status}) from ${url}${text ? `: ${text}` : ""}`);
+  }
+  const json = await res.json();
+  if (!isPresetPayloadV1(json)) {
+    throw new Error(`Invalid preset payload returned from ${url}`);
+  }
+  return json;
+}
+
+// cli/lib/packageJson.ts
+import path4 from "node:path";
+import { readFile as readFile4, writeFile as writeFile3 } from "node:fs/promises";
 async function detectPackageManager(projectRoot) {
-  if (await exists(path3.join(projectRoot, "bun.lock")))
+  if (await exists(path4.join(projectRoot, "bun.lock")))
     return "bun";
-  if (await exists(path3.join(projectRoot, "pnpm-lock.yaml")))
+  if (await exists(path4.join(projectRoot, "pnpm-lock.yaml")))
     return "pnpm";
-  if (await exists(path3.join(projectRoot, "yarn.lock")))
+  if (await exists(path4.join(projectRoot, "yarn.lock")))
     return "yarn";
-  if (await exists(path3.join(projectRoot, "package-lock.json")))
+  if (await exists(path4.join(projectRoot, "package-lock.json")))
     return "npm";
   return "npm";
 }
 async function readPackageJson(projectRoot) {
-  const raw = await readFile3(path3.join(projectRoot, "package.json"), "utf8");
+  const raw = await readFile4(path4.join(projectRoot, "package.json"), "utf8");
   return JSON.parse(raw);
 }
 async function writePackageJson(projectRoot, pkg) {
-  await writeFile3(path3.join(projectRoot, "package.json"), JSON.stringify(pkg, null, 2) + `
+  await writeFile3(path4.join(projectRoot, "package.json"), JSON.stringify(pkg, null, 2) + `
 `, "utf8");
 }
 function mergeDeps(pkg, deps, devDeps) {
@@ -431,7 +715,7 @@ function mergeDeps(pkg, deps, devDeps) {
 }
 function depsForAdapter(adapter) {
   const deps = {
-    ai: "^5.0.113"
+    ai: "^6.0.3"
   };
   const devDeps = {};
   if (adapter === "drizzle") {
@@ -465,20 +749,20 @@ function depsForExtractor(extractor) {
   const deps = {};
   const devDeps = {};
   if (extractor === "pdf-llm") {
-    deps["ai"] = "^5.0.113";
+    deps["ai"] = "^6.0.3";
   }
   if (extractor === "pdf-text-layer") {
     deps["pdfjs-dist"] = "^5.4.149";
   }
   if (extractor === "pdf-ocr") {}
   if (extractor === "image-ocr" || extractor === "image-caption-llm") {
-    deps["ai"] = "^5.0.113";
+    deps["ai"] = "^6.0.3";
   }
   if (extractor === "audio-transcribe" || extractor === "video-transcribe") {
-    deps["ai"] = "^5.0.113";
+    deps["ai"] = "^6.0.3";
   }
   if (extractor === "video-frames") {
-    deps["ai"] = "^5.0.113";
+    deps["ai"] = "^6.0.3";
   }
   if (extractor === "file-text") {}
   if (extractor === "file-docx") {
@@ -492,6 +776,33 @@ function depsForExtractor(extractor) {
   }
   return { deps, devDeps };
 }
+function depsForEmbeddingProvider(provider) {
+  const deps = {};
+  const devDeps = {};
+  if (provider === "openai")
+    deps["@ai-sdk/openai"] = "^3.0.1";
+  if (provider === "google")
+    deps["@ai-sdk/google"] = "^3.0.1";
+  if (provider === "azure")
+    deps["@ai-sdk/azure"] = "^3.0.1";
+  if (provider === "vertex")
+    deps["@ai-sdk/google-vertex"] = "^3.0.1";
+  if (provider === "bedrock")
+    deps["@ai-sdk/amazon-bedrock"] = "^3.0.72";
+  if (provider === "cohere")
+    deps["@ai-sdk/cohere"] = "^3.0.1";
+  if (provider === "mistral")
+    deps["@ai-sdk/mistral"] = "^3.0.1";
+  if (provider === "together")
+    deps["@ai-sdk/togetherai"] = "^3.0.1";
+  if (provider === "openrouter")
+    deps["@openrouter/sdk"] = "^0.3.10";
+  if (provider === "ollama")
+    deps["ollama-ai-provider-v2"] = "^2.0.0";
+  if (provider === "voyage")
+    deps["voyage-ai-provider"] = "^3.0.0";
+  return { deps, devDeps };
+}
 function installCmd(pm) {
   if (pm === "bun")
     return "bun install";
@@ -503,8 +814,8 @@ function installCmd(pm) {
 }
 
 // cli/lib/tsconfig.ts
-import path4 from "node:path";
-import { readFile as readFile4, writeFile as writeFile4 } from "node:fs/promises";
+import path5 from "node:path";
+import { readFile as readFile5, writeFile as writeFile4 } from "node:fs/promises";
 import { parse } from "jsonc-parser";
 var parseJsoncLoose = (raw) => {
   const errors = [];
@@ -515,14 +826,14 @@ var parseJsoncLoose = (raw) => {
   return result;
 };
 async function patchTsconfigPaths(params) {
-  const configFile = await exists(path4.join(params.projectRoot, "tsconfig.json")) ? "tsconfig.json" : await exists(path4.join(params.projectRoot, "jsconfig.json")) ? "jsconfig.json" : null;
+  const configFile = await exists(path5.join(params.projectRoot, "tsconfig.json")) ? "tsconfig.json" : await exists(path5.join(params.projectRoot, "jsconfig.json")) ? "jsconfig.json" : null;
   const aliasBase = params.aliasBase;
   const aliasKey = `${aliasBase}/*`;
   const target = [`./${params.installDir.replace(/\\/g, "/")}/*`];
   const configAliasKey = `${aliasBase}/config`;
   const configTarget = ["./unrag.config.ts"];
   if (!configFile) {
-    const abs2 = path4.join(params.projectRoot, "tsconfig.json");
+    const abs2 = path5.join(params.projectRoot, "tsconfig.json");
     const next2 = {
       compilerOptions: {
         baseUrl: ".",
@@ -536,8 +847,8 @@ async function patchTsconfigPaths(params) {
 `, "utf8");
     return { changed: true, file: "tsconfig.json" };
   }
-  const abs = path4.join(params.projectRoot, configFile);
-  const raw = await readFile4(abs, "utf8");
+  const abs = path5.join(params.projectRoot, configFile);
+  const raw = await readFile5(abs, "utf8");
   let parsed;
   try {
     parsed = parseJsoncLoose(raw);
@@ -572,7 +883,7 @@ async function patchTsconfigPaths(params) {
 var CONFIG_FILE = "unrag.json";
 var CONFIG_VERSION = 1;
 var __filename2 = fileURLToPath(import.meta.url);
-var __dirname2 = path5.dirname(__filename2);
+var __dirname2 = path6.dirname(__filename2);
 var parseInitArgs = (args) => {
   const out = {};
   for (let i = 0;i < args.length; i++) {
@@ -621,77 +932,35 @@ var parseInitArgs = (args) => {
       }
       continue;
     }
+    if (a === "--provider") {
+      const v = args[i + 1];
+      if (v === "ai" || v === "openai" || v === "google" || v === "openrouter" || v === "azure" || v === "vertex" || v === "bedrock" || v === "cohere" || v === "mistral" || v === "together" || v === "ollama" || v === "voyage") {
+        out.provider = v;
+        i++;
+      }
+      continue;
+    }
+    if (a === "--preset") {
+      const v = args[i + 1];
+      if (v) {
+        out.preset = v;
+        i++;
+      }
+      continue;
+    }
+    if (a === "--overwrite") {
+      const v = args[i + 1];
+      if (v === "skip" || v === "force") {
+        out.overwrite = v;
+        i++;
+      }
+      continue;
+    }
   }
   return out;
 };
-var DEFAULT_RICH_MEDIA_EXTRACTORS = ["pdf-text-layer", "file-text"];
-var EXTRACTOR_OPTIONS = [
-  {
-    group: "PDF",
-    value: "pdf-text-layer",
-    label: `pdf-text-layer (Fast/cheap extraction via PDF text layer)`,
-    hint: "recommended"
-  },
-  {
-    group: "PDF",
-    value: "pdf-llm",
-    label: `pdf-llm (LLM-based PDF extraction; higher cost)`
-  },
-  {
-    group: "PDF",
-    value: "pdf-ocr",
-    label: `pdf-ocr (OCR scanned PDFs; requires native binaries)`,
-    hint: "worker-only"
-  },
-  {
-    group: "Image",
-    value: "image-ocr",
-    label: `image-ocr (Extract text from images via vision LLM)`
-  },
-  {
-    group: "Image",
-    value: "image-caption-llm",
-    label: `image-caption-llm (Generate captions for images via vision LLM)`
-  },
-  {
-    group: "Audio",
-    value: "audio-transcribe",
-    label: `audio-transcribe (Speech-to-text transcription)`
-  },
-  {
-    group: "Video",
-    value: "video-transcribe",
-    label: `video-transcribe (Transcribe video audio track)`
-  },
-  {
-    group: "Video",
-    value: "video-frames",
-    label: `video-frames (Sample frames + analyze via vision LLM; requires ffmpeg)`,
-    hint: "worker-only"
-  },
-  {
-    group: "Files",
-    value: "file-text",
-    label: `file-text (Extract text/markdown/json/html from common text files)`,
-    hint: "recommended"
-  },
-  {
-    group: "Files",
-    value: "file-docx",
-    label: `file-docx (Extract text from .docx files)`
-  },
-  {
-    group: "Files",
-    value: "file-pptx",
-    label: `file-pptx (Extract text from .pptx slides)`
-  },
-  {
-    group: "Files",
-    value: "file-xlsx",
-    label: `file-xlsx (Extract tables from .xlsx spreadsheets)`
-  }
-];
-var AVAILABLE_EXTRACTORS = new Set(EXTRACTOR_OPTIONS.map((o) => o.value));
+var toExtractors = (xs) => (Array.isArray(xs) ? xs : []).map((s) => String(s).trim()).filter(Boolean);
+var toConnectors = (xs) => (Array.isArray(xs) ? xs : []).map((s) => String(s).trim()).filter(Boolean);
 async function initCommand(args) {
   const root = await tryFindProjectRoot(process.cwd());
   if (!root) {
@@ -701,15 +970,42 @@ async function initCommand(args) {
   if (!cliPackageRoot) {
     throw new Error("Could not locate CLI package root (package.json not found).");
   }
-  const registryRoot = path5.join(cliPackageRoot, "registry");
-  const existing = await readJsonFile(path5.join(root, CONFIG_FILE));
+  const registryRoot = path6.join(cliPackageRoot, "registry");
+  const manifest = await readRegistryManifest(registryRoot);
+  const extractorOptions = manifest.extractors.map((ex) => {
+    const value = ex.id;
+    const label = ex.description ? `${ex.label} (${ex.description})` : ex.label;
+    return {
+      group: ex.group,
+      value,
+      label,
+      hint: ex.hint,
+      defaultSelected: Boolean(ex.defaultSelected)
+    };
+  });
+  const availableExtractors = new Set(extractorOptions.map((o) => o.value));
+  const defaultRichMediaExtractors = extractorOptions.filter((o) => o.defaultSelected).map((o) => o.value).sort();
+  const existing = await readJsonFile(path6.join(root, CONFIG_FILE));
   const parsed = parseInitArgs(args);
+  const preset = parsed.preset ? await fetchPreset(parsed.preset) : null;
+  if (preset) {
+    const hasOtherChoices = Boolean(parsed.installDir) || Boolean(parsed.storeAdapter) || Boolean(parsed.aliasBase) || typeof parsed.richMedia === "boolean" || (parsed.extractors ?? []).length > 0;
+    if (hasOtherChoices) {
+      throw new Error('When using "--preset", do not pass other init preference flags (--store/--dir/--alias/--rich-media/--extractors).');
+    }
+  }
+  const presetEmbeddingProvider = (() => {
+    const v = preset?.config?.embedding?.provider;
+    return v === "ai" || v === "openai" || v === "google" || v === "openrouter" || v === "azure" || v === "vertex" || v === "bedrock" || v === "cohere" || v === "mistral" || v === "together" || v === "ollama" || v === "voyage" || v === "custom" ? v : undefined;
+  })();
   const defaults = {
-    installDir: existing?.installDir ?? "lib/unrag",
-    storeAdapter: existing?.storeAdapter ?? "drizzle",
-    aliasBase: existing?.aliasBase ?? "@unrag"
+    installDir: preset?.install?.installDir ?? existing?.installDir ?? "lib/unrag",
+    storeAdapter: preset?.install?.storeAdapter ?? existing?.storeAdapter ?? "drizzle",
+    aliasBase: preset?.install?.aliasBase ?? existing?.aliasBase ?? "@unrag",
+    embeddingProvider: parsed.provider ?? presetEmbeddingProvider ?? existing?.embeddingProvider ?? "ai"
   };
-  const nonInteractive = parsed.yes || !process.stdin.isTTY;
+  const nonInteractive = Boolean(parsed.yes) || Boolean(preset) || !process.stdin.isTTY;
+  const overwritePolicy = parsed.overwrite ?? "skip";
   const installDirAnswer = parsed.installDir ? parsed.installDir : nonInteractive ? defaults.installDir : await text({
     message: "Install directory",
     initialValue: defaults.installDir,
@@ -760,12 +1056,41 @@ async function initCommand(args) {
     return;
   }
   const aliasBase = String(aliasAnswer).trim();
+  const embeddingProviderAnswer = parsed.provider ? parsed.provider : nonInteractive ? defaults.embeddingProvider : await select({
+    message: "Embedding provider",
+    initialValue: defaults.embeddingProvider,
+    options: [
+      { value: "ai", label: "Vercel AI Gateway (AI SDK)", hint: "default" },
+      { value: "openai", label: "OpenAI" },
+      { value: "google", label: "Google AI (Gemini)" },
+      { value: "openrouter", label: "OpenRouter" },
+      { value: "azure", label: "Azure OpenAI" },
+      { value: "vertex", label: "Google Vertex AI" },
+      { value: "bedrock", label: "AWS Bedrock" },
+      { value: "cohere", label: "Cohere" },
+      { value: "mistral", label: "Mistral" },
+      { value: "together", label: "Together.ai" },
+      { value: "ollama", label: "Ollama (local)" },
+      { value: "voyage", label: "Voyage AI" }
+    ]
+  });
+  if (isCancel2(embeddingProviderAnswer)) {
+    cancel2("Cancelled.");
+    return;
+  }
+  const embeddingProvider = embeddingProviderAnswer;
   if (parsed.richMedia === false && (parsed.extractors ?? []).length > 0) {
     throw new Error('Cannot use "--no-rich-media" together with "--extractors".');
   }
-  const extractorsFromArgs = (parsed.extractors ?? []).filter((x) => AVAILABLE_EXTRACTORS.has(x)).sort();
+  const extractorsFromArgs = (preset ? toExtractors(preset.modules?.extractors) : parsed.extractors ?? []).filter((x) => availableExtractors.has(x)).sort();
+  if (preset) {
+    const unknown = toExtractors(preset.modules?.extractors).filter((x) => !availableExtractors.has(x));
+    if (unknown.length > 0) {
+      throw new Error(`Preset contains unknown extractors: ${unknown.join(", ")}`);
+    }
+  }
   const richMediaAnswer = extractorsFromArgs.length > 0 ? true : typeof parsed.richMedia === "boolean" ? parsed.richMedia : nonInteractive ? false : await confirm2({
-    message: "Enable rich media ingestion (PDF/images/audio/video/files)? This also enables multimodal image embeddings (you can change this later).",
+    message: "Enable rich media ingestion (PDF/images/audio/video/files)? This enables extractor modules and assetProcessing (you can change this later).",
     initialValue: false
   });
   if (isCancel2(richMediaAnswer)) {
@@ -773,9 +1098,9 @@ async function initCommand(args) {
     return;
   }
   const richMediaEnabled = Boolean(richMediaAnswer);
-  const selectedExtractorsAnswer = richMediaEnabled || extractorsFromArgs.length > 0 ? nonInteractive ? extractorsFromArgs.length > 0 ? extractorsFromArgs : DEFAULT_RICH_MEDIA_EXTRACTORS : await groupMultiselect({
+  const selectedExtractorsAnswer = richMediaEnabled || extractorsFromArgs.length > 0 ? nonInteractive ? extractorsFromArgs.length > 0 ? extractorsFromArgs : defaultRichMediaExtractors.length > 0 ? defaultRichMediaExtractors : ["pdf-text-layer", "file-text"] : await groupMultiselect({
     message: "Select extractors to enable (space to toggle, enter to confirm)",
-    options: EXTRACTOR_OPTIONS.reduce((acc, opt) => {
+    options: extractorOptions.reduce((acc, opt) => {
       acc[opt.group] ??= [];
       acc[opt.group].push({
         value: opt.value,
@@ -784,7 +1109,7 @@ async function initCommand(args) {
       });
       return acc;
     }, {}),
-    initialValues: extractorsFromArgs.length > 0 ? extractorsFromArgs : DEFAULT_RICH_MEDIA_EXTRACTORS,
+    initialValues: extractorsFromArgs.length > 0 ? extractorsFromArgs : defaultRichMediaExtractors.length > 0 ? defaultRichMediaExtractors : ["pdf-text-layer", "file-text"],
     required: false
   }) : [];
   if (isCancel2(selectedExtractorsAnswer)) {
@@ -798,6 +1123,10 @@ async function initCommand(args) {
     projectRoot: root,
     registryRoot,
     aliasBase,
+    embeddingProvider,
+    yes: nonInteractive,
+    overwrite: overwritePolicy,
+    presetConfig: preset?.config ?? undefined,
     richMedia: richMediaEnabled ? {
       enabled: true,
       extractors: selectedExtractors
@@ -811,12 +1140,14 @@ async function initCommand(args) {
         registryRoot,
         installDir,
         extractor,
-        yes: nonInteractive
+        yes: nonInteractive,
+        overwrite: overwritePolicy
       });
     }
   }
   const pkg = await readPackageJson(root);
   const { deps, devDeps } = depsForAdapter(storeAdapterAnswer);
+  const embeddingDeps = depsForEmbeddingProvider(embeddingProvider);
   const extractorDeps = {};
   const extractorDevDeps = {};
   for (const ex of selectedExtractors) {
@@ -824,7 +1155,34 @@ async function initCommand(args) {
     Object.assign(extractorDeps, r.deps);
     Object.assign(extractorDevDeps, r.devDeps);
   }
-  const merged = mergeDeps(pkg, { ...deps, ...extractorDeps }, { ...devDeps, ...extractorDevDeps });
+  const connectorsFromPreset = preset ? toConnectors(preset.modules?.connectors) : [];
+  const availableConnectorIds = new Set((manifest.connectors ?? []).filter((c) => c.status === "available").map((c) => String(c.id)));
+  if (preset) {
+    const unknown = connectorsFromPreset.filter((c) => !availableConnectorIds.has(c));
+    if (unknown.length > 0) {
+      throw new Error(`Preset contains unknown/unavailable connectors: ${unknown.join(", ")}`);
+    }
+  }
+  if (connectorsFromPreset.length > 0) {
+    for (const connector of connectorsFromPreset) {
+      await copyConnectorFiles({
+        projectRoot: root,
+        registryRoot,
+        installDir,
+        connector,
+        yes: nonInteractive,
+        overwrite: overwritePolicy
+      });
+    }
+  }
+  const connectorDeps = {};
+  const connectorDevDeps = {};
+  for (const c of connectorsFromPreset) {
+    const r = depsForConnector(c);
+    Object.assign(connectorDeps, r.deps);
+    Object.assign(connectorDevDeps, r.devDeps);
+  }
+  const merged = mergeDeps(pkg, { ...deps, ...embeddingDeps.deps, ...extractorDeps, ...connectorDeps }, { ...devDeps, ...embeddingDeps.devDeps, ...extractorDevDeps, ...connectorDevDeps });
   if (merged.changes.length > 0) {
     await writePackageJson(root, merged.pkg);
   }
@@ -832,28 +1190,126 @@ async function initCommand(args) {
     installDir,
     storeAdapter: storeAdapterAnswer,
     aliasBase,
+    embeddingProvider,
     version: CONFIG_VERSION,
-    connectors: existing?.connectors ?? [],
+    connectors: Array.from(new Set([...existing?.connectors ?? [], ...connectorsFromPreset])).sort(),
     extractors: Array.from(new Set([
       ...existing?.extractors ?? [],
       ...richMediaEnabled ? selectedExtractors : []
     ])).sort()
   };
-  await writeJsonFile(path5.join(root, CONFIG_FILE), config);
+  await writeJsonFile(path6.join(root, CONFIG_FILE), config);
   const pm = await detectPackageManager(root);
   const installLine = merged.changes.length > 0 ? `Next: run \`${installCmd(pm)}\`` : "Dependencies already satisfied.";
   const isNext = Boolean((merged.pkg.dependencies ?? {})["next"]) || Boolean((merged.pkg.devDependencies ?? {})["next"]);
   const tsconfigResult = isNext ? await patchTsconfigPaths({ projectRoot: root, installDir, aliasBase }) : { changed: false };
+  const envHint = (() => {
+    if (embeddingProvider === "ai") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- AI_GATEWAY_API_KEY=...",
+        "- (optional) AI_GATEWAY_MODEL=openai/text-embedding-3-small"
+      ];
+    }
+    if (embeddingProvider === "openai") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- OPENAI_API_KEY=...",
+        "- (optional) OPENAI_EMBEDDING_MODEL=text-embedding-3-small"
+      ];
+    }
+    if (embeddingProvider === "google") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- GOOGLE_GENERATIVE_AI_API_KEY=...",
+        "- (optional) GOOGLE_GENERATIVE_AI_EMBEDDING_MODEL=gemini-embedding-001"
+      ];
+    }
+    if (embeddingProvider === "openrouter") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- OPENROUTER_API_KEY=...",
+        "- (optional) OPENROUTER_EMBEDDING_MODEL=text-embedding-3-small"
+      ];
+    }
+    if (embeddingProvider === "cohere") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- COHERE_API_KEY=...",
+        "- (optional) COHERE_EMBEDDING_MODEL=embed-english-v3.0"
+      ];
+    }
+    if (embeddingProvider === "mistral") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- MISTRAL_API_KEY=...",
+        "- (optional) MISTRAL_EMBEDDING_MODEL=mistral-embed"
+      ];
+    }
+    if (embeddingProvider === "together") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- TOGETHER_AI_API_KEY=...",
+        "- (optional) TOGETHER_AI_EMBEDDING_MODEL=togethercomputer/m2-bert-80M-2k-retrieval"
+      ];
+    }
+    if (embeddingProvider === "voyage") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- VOYAGE_API_KEY=...",
+        "- (optional) VOYAGE_MODEL=voyage-3.5-lite"
+      ];
+    }
+    if (embeddingProvider === "ollama") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- (optional) OLLAMA_EMBEDDING_MODEL=nomic-embed-text"
+      ];
+    }
+    if (embeddingProvider === "azure") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- AZURE_OPENAI_API_KEY=...",
+        "- AZURE_RESOURCE_NAME=...",
+        "- (optional) AZURE_EMBEDDING_MODEL=text-embedding-3-small"
+      ];
+    }
+    if (embeddingProvider === "vertex") {
+      return [
+        "Env:",
+        "- DATABASE_URL=...",
+        "- GOOGLE_APPLICATION_CREDENTIALS=... (when outside GCP)",
+        "- (optional) GOOGLE_VERTEX_EMBEDDING_MODEL=text-embedding-004"
+      ];
+    }
+    return [
+      "Env:",
+      "- DATABASE_URL=...",
+      "- AWS_REGION=... (Bedrock)",
+      "- AWS credentials (when outside AWS)",
+      "- (optional) BEDROCK_EMBEDDING_MODEL=amazon.titan-embed-text-v2:0"
+    ];
+  })();
   outro([
     "Installed Unrag.",
     "",
-    `- Code: ${path5.join(installDir)}`,
-    `- Docs: ${path5.join(installDir, "unrag.md")}`,
+    `- Code: ${path6.join(installDir)}`,
+    `- Docs: ${path6.join(installDir, "unrag.md")}`,
     `- Config: unrag.config.ts`,
     `- Imports: ${aliasBase}/* and ${aliasBase}/config`,
     "",
     `- Rich media: ${richMediaEnabled ? "enabled" : "disabled"}`,
-    richMediaEnabled ? `- Embeddings: multimodal enabled (images can be embedded directly)` : `- Embeddings: text-only (no direct image embedding)`,
+    `- Embedding provider: ${embeddingProvider}`,
     richMediaEnabled ? `- Extractors: ${selectedExtractors.length > 0 ? selectedExtractors.join(", ") : "none"}` : "",
     richMediaEnabled ? `  Tip: you can tweak extractors + assetProcessing flags in unrag.config.ts later.` : `  Tip: re-run \`unrag init --rich-media\` (or edit unrag.config.ts) to enable rich media later.`,
     isNext ? tsconfigResult.changed ? `- Next.js: updated ${tsconfigResult.file} (added aliases)` : `- Next.js: no tsconfig changes needed` : `- Next.js: not detected`,
@@ -861,6 +1317,8 @@ async function initCommand(args) {
     merged.changes.length > 0 ? `Added deps: ${merged.changes.map((c) => c.name).join(", ")}` : "Added deps: none",
     installLine,
     "",
+    ...envHint,
+    "",
     `Saved ${CONFIG_FILE}.`
   ].join(`
 `));
@@ -868,37 +1326,11 @@ async function initCommand(args) {
 
 // cli/commands/add.ts
 import { outro as outro2 } from "@clack/prompts";
-import path6 from "node:path";
+import path7 from "node:path";
 import { fileURLToPath as fileURLToPath2 } from "node:url";
-
-// cli/lib/constants.ts
-var UNRAG_SITE_URL = (process.env.UNRAG_SITE_URL ?? process.env.UNRAG_DOCS_BASE_URL)?.trim() || "https://unrag.dev";
-var UNRAG_GITHUB_REPO_URL = "https://github.com/BetterStacks/unrag";
-function docsUrl(siteRelativePath) {
-  const p = siteRelativePath.startsWith("/") ? siteRelativePath : `/${siteRelativePath}`;
-  const base = UNRAG_SITE_URL.endsWith("/") ? UNRAG_SITE_URL : `${UNRAG_SITE_URL}/`;
-  return new URL(p.replace(/^\/+/, "/"), base).toString();
-}
-
-// cli/commands/add.ts
 var CONFIG_FILE2 = "unrag.json";
 var __filename3 = fileURLToPath2(import.meta.url);
-var __dirname3 = path6.dirname(__filename3);
-var AVAILABLE_EXTRACTORS2 = [
-  "pdf-llm",
-  "pdf-text-layer",
-  "pdf-ocr",
-  "image-ocr",
-  "image-caption-llm",
-  "audio-transcribe",
-  "video-transcribe",
-  "video-frames",
-  "file-text",
-  "file-docx",
-  "file-pptx",
-  "file-xlsx"
-];
-var AVAILABLE_CONNECTORS = ["notion", "google-drive"];
+var __dirname3 = path7.dirname(__filename3);
 var parseAddArgs = (args) => {
   const out = {};
   for (let i = 0;i < args.length; i++) {
@@ -931,36 +1363,39 @@ async function addCommand(args) {
   const parsed = parseAddArgs(args);
   const kind = parsed.kind ?? "connector";
   const name = parsed.name;
+  const configPath = path7.join(root, CONFIG_FILE2);
+  const config = await readJsonFile(configPath);
+  if (!config?.installDir) {
+    throw new Error(`Missing ${CONFIG_FILE2}. Run \`unrag@latest init\` first.`);
+  }
+  const cliPackageRoot = await findUp(__dirname3, "package.json");
+  if (!cliPackageRoot) {
+    throw new Error("Could not locate CLI package root (package.json not found).");
+  }
+  const registryRoot = path7.join(cliPackageRoot, "registry");
+  const manifest = await readRegistryManifest(registryRoot);
+  const availableExtractors = new Set(manifest.extractors.map((e) => e.id));
+  const availableConnectors = new Set(manifest.connectors.filter((c) => c.status === "available").map((c) => c.id));
   if (!name) {
     outro2([
       "Usage:",
       "  unrag add <connector>",
       "  unrag add extractor <name>",
       "",
-      `Available connectors: ${AVAILABLE_CONNECTORS.join(", ")}`,
-      `Available extractors: ${AVAILABLE_EXTRACTORS2.join(", ")}`
+      `Available connectors: ${Array.from(availableConnectors).join(", ")}`,
+      `Available extractors: ${Array.from(availableExtractors).join(", ")}`
     ].join(`
 `));
     return;
   }
-  const configPath = path6.join(root, CONFIG_FILE2);
-  const config = await readJsonFile(configPath);
-  if (!config?.installDir) {
-    throw new Error(`Missing ${CONFIG_FILE2}. Run \`unrag@latest init\` first.`);
-  }
-  const cliPackageRoot = await findUp(__dirname3, "package.json");
-  if (!cliPackageRoot) {
-    throw new Error("Could not locate CLI package root (package.json not found).");
-  }
-  const registryRoot = path6.join(cliPackageRoot, "registry");
   const nonInteractive = parsed.yes || !process.stdin.isTTY;
   const pkg = await readPackageJson(root);
   if (kind === "connector") {
     const connector = name;
-    if (!connector || !AVAILABLE_CONNECTORS.includes(connector)) {
+    if (!connector || !availableConnectors.has(connector)) {
       outro2(`Unknown connector: ${name}
 
-Available connectors: ${AVAILABLE_CONNECTORS.join(", ")}`);
+Available connectors: ${Array.from(availableConnectors).join(", ")}`);
       return;
     }
     await copyConnectorFiles({
@@ -980,7 +1415,7 @@ Available connectors: ${AVAILABLE_CONNECTORS.join(", ")}`);
     outro2([
       `Installed connector: ${connector}.`,
       "",
-      `- Code: ${path6.join(config.installDir, "connectors", connector)}`,
+      `- Code: ${path7.join(config.installDir, "connectors", connector)}`,
       `- Docs: ${docsUrl(`/docs/connectors/${connector}`)}`,
       "",
       merged2.changes.length > 0 ? `Added deps: ${merged2.changes.map((c) => c.name).join(", ")}` : "Added deps: none",
@@ -990,10 +1425,10 @@ Available connectors: ${AVAILABLE_CONNECTORS.join(", ")}`);
     return;
   }
   const extractor = name;
-  if (!extractor || !AVAILABLE_EXTRACTORS2.includes(extractor)) {
+  if (!extractor || !availableExtractors.has(extractor)) {
     outro2(`Unknown extractor: ${name}
 
-Available extractors: ${AVAILABLE_EXTRACTORS2.join(", ")}`);
+Available extractors: ${Array.from(availableExtractors).join(", ")}`);
     return;
   }
   await copyExtractorFiles({
@@ -1013,7 +1448,7 @@ Available extractors: ${AVAILABLE_EXTRACTORS2.join(", ")}`);
   outro2([
     `Installed extractor: ${extractor}.`,
     "",
-    `- Code: ${path6.join(config.installDir, "extractors", extractor)}`,
+    `- Code: ${path7.join(config.installDir, "extractors", extractor)}`,
     "",
     merged.changes.length > 0 ? `Added deps: ${merged.changes.map((c) => c.name).join(", ")}` : "Added deps: none",
     "",
@@ -1033,7 +1468,7 @@ function renderHelp() {
     "",
     "Commands:",
     "  init                Install core files (config + store adapter templates)",
-    "  add <connector>     Install a connector (currently: notion)",
+    "  add <connector>     Install a connector (notion, google-drive)",
     "  help                Show this help",
     "",
     "Global options:",
@@ -1044,6 +1479,8 @@ function renderHelp() {
     "  --store <adapter>    drizzle | prisma | raw-sql",
     "  --dir <path>         Install directory (alias: --install-dir)",
     "  --alias <@name>      Import alias base (e.g. @unrag)",
+    "  --preset <id|url>    Install from a web-generated preset (non-interactive)",
+    "  --overwrite <mode>   skip | force (when files already exist)",
     "  --rich-media         Enable rich media setup (also enables multimodal embeddings)",
     "  --no-rich-media      Disable rich media setup",
     "  --extractors <list>  Comma-separated extractors (implies --rich-media)",