@open-mercato/web-research-model 0.7.1-develop.7194.1.ab4fc81f82

package/.turbo/turbo-build.log

@@ -0,0 +1,2 @@
+[build:web-research-model] found 4 entry points
+[build:web-research-model] built successfully

package/build.mjs

@@ -0,0 +1,10 @@
+import { dirname } from 'node:path'
+import { fileURLToPath } from 'node:url'
+import { buildPackage } from '../../scripts/build-package.mjs'
+
+const packageDir = dirname(fileURLToPath(import.meta.url))
+
+await buildPackage(packageDir, {
+  name: 'web-research-model',
+  entryPoints: 'src/**/*.ts',
+})

package/dist/adapter.js

@@ -0,0 +1,173 @@
+import {
+  CONTRACT_VERSION,
+  READY,
+  defineAdapterModule,
+  notReady,
+  searchOk
+} from "@open-mercato/web-research";
+import { z } from "zod";
+import { loadNativeSearchTool } from "./nativeTools.js";
+import { parseModelSummaries, stripSummaryBlock } from "./parse.js";
+const modelOptionsSchema = z.object({
+  maxUses: z.number().int().min(1).max(10).optional(),
+  resolveModel: z.custom((value) => typeof value === "function").optional(),
+  generate: z.custom((value) => typeof value === "function").optional(),
+  loadTools: z.custom((value) => typeof value === "function").optional()
+});
+const DEFAULT_MAX_USES = 3;
+const NOT_CONFIGURED_MARKERS = [
+  "exceeded your current quota",
+  "insufficient_quota",
+  "billing",
+  "payment required",
+  "invalid api key",
+  "incorrect api key",
+  "invalid_api_key",
+  "authentication",
+  "unauthorized",
+  "account is not active"
+];
+const PROVIDER_CAVEATS = {
+  azure: "searches run via Grounding with Bing, billed separately and outside your geo boundary"
+};
+function isConfigurationFault(message) {
+  const haystack = message.toLowerCase();
+  return NOT_CONFIGURED_MARKERS.some((marker) => haystack.includes(marker));
+}
+function buildPrompt(request) {
+  const constraints = [];
+  if (request.site) constraints.push(`Restrict results to the site ${request.site}.`);
+  if (request.freshness) constraints.push(`Prefer sources published within the last ${request.freshness}.`);
+  if (request.locale) constraints.push(`Prefer sources written for the ${request.locale} locale.`);
+  return [
+    `Search the public web for: ${request.query}`,
+    "Use the web_search tool, then answer in two parts.",
+    "",
+    "Part 1: a short prose answer citing the most relevant, authoritative sources.",
+    "",
+    `Part 2: a JSON array in a \`\`\`json code fence with up to ${request.limit} entries, each`,
+    '{"url": "...", "title": "...", "summary": "one or two sentences on what this page says about the query"}.',
+    "Only include URLs you actually retrieved. Write real summaries, not the page title again.",
+    ...constraints
+  ].join("\n");
+}
+async function defaultGenerate(input) {
+  const { generateText } = await import("ai");
+  return await generateText({
+    model: input.model,
+    tools: input.tools,
+    prompt: input.prompt,
+    ...input.abortSignal ? { abortSignal: input.abortSignal } : {}
+  });
+}
+function createModelAdapter(options) {
+  const maxUses = options.maxUses ?? DEFAULT_MAX_USES;
+  const resolveModel = options.resolveModel;
+  const generate = options.generate ?? defaultGenerate;
+  const loadTools = options.loadTools ?? loadNativeSearchTool;
+  return {
+    id: "model-native",
+    kind: "model",
+    capabilities: {
+      search: true,
+      fetch: false,
+      snippets: true,
+      content: false,
+      freshness: true,
+      siteFilter: true,
+      cost: "llm-tokens"
+    },
+    /** `free` — resolves the configured model and its tool descriptors; no completion is issued. */
+    probeCost: "free",
+    readiness: () => resolveModel ? READY : notReady("no model resolver was provided by the host application"),
+    async search(request, context) {
+      if (!resolveModel) {
+        return { status: "unavailable", reason: "no model resolver was provided by the host application" };
+      }
+      let resolution;
+      try {
+        resolution = await resolveModel();
+      } catch (error) {
+        return {
+          status: "unavailable",
+          reason: `could not resolve a model: ${error instanceof Error ? error.message : String(error)}`
+        };
+      }
+      const tools = await loadTools(resolution.providerId, maxUses);
+      if (!tools) {
+        return {
+          status: "unavailable",
+          reason: `provider "${resolution.providerId}" has no native web search available`
+        };
+      }
+      let result;
+      try {
+        result = await generate({
+          model: resolution.model,
+          tools,
+          prompt: buildPrompt(request),
+          abortSignal: context.signal
+        });
+      } catch (error) {
+        if (context.signal.aborted) return { status: "timeout", reason: "model search was aborted" };
+        const detail = error instanceof Error ? error.message : String(error);
+        if (isConfigurationFault(detail)) {
+          return { status: "unavailable", reason: `model search is not usable: ${detail}` };
+        }
+        return { status: "error", reason: `model search failed: ${detail}`, retriable: true };
+      }
+      const text = result.text ?? "";
+      const summaries = new Map(parseModelSummaries(text).map((entry) => [entry.url, entry]));
+      const results = [];
+      const seen = /* @__PURE__ */ new Set();
+      for (const source of result.sources ?? []) {
+        if (source.sourceType && source.sourceType !== "url") continue;
+        const url = source.url;
+        if (!url || seen.has(url)) continue;
+        seen.add(url);
+        const enrichment = summaries.get(url);
+        results.push({
+          url,
+          title: enrichment?.title ?? source.title ?? null,
+          snippet: enrichment?.summary ?? null
+        });
+        if (results.length >= request.limit) break;
+      }
+      const answer = stripSummaryBlock(text);
+      context.report("ok", `model returned ${results.length} cited source(s)`, {
+        resultCount: results.length
+      });
+      return searchOk(results, answer);
+    },
+    async healthCheck() {
+      if (!resolveModel) return { ok: false, detail: "no model resolver was provided by the host application" };
+      try {
+        const resolution = await resolveModel();
+        const tools = await loadTools(resolution.providerId, maxUses);
+        if (!tools) {
+          return { ok: false, detail: `provider "${resolution.providerId}" has no native web search` };
+        }
+        const caveat = PROVIDER_CAVEATS[resolution.providerId];
+        return {
+          ok: true,
+          detail: `${resolution.providerId}/${resolution.modelId}${caveat ? ` - ${caveat}` : ""}`
+        };
+      } catch (error) {
+        return { ok: false, detail: error instanceof Error ? error.message : String(error) };
+      }
+    }
+  };
+}
+const modelAdapterModule = defineAdapterModule({
+  id: "model-native",
+  kind: "model",
+  contractVersion: CONTRACT_VERSION,
+  optionsSchema: modelOptionsSchema,
+  createAdapter: createModelAdapter
+});
+export {
+  createModelAdapter,
+  modelAdapterModule,
+  modelOptionsSchema
+};
+//# sourceMappingURL=adapter.js.map

package/dist/adapter.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/adapter.ts"],
+  "sourcesContent": ["import {\n  CONTRACT_VERSION,\n  READY,\n  defineAdapterModule,\n  notReady,\n  searchOk,\n  type RawResult,\n  type SearchAdapter,\n  type SearchOutcome,\n  type SearchRequest,\n} from '@open-mercato/web-research'\nimport { z } from 'zod'\nimport { loadNativeSearchTool } from './nativeTools'\nimport { parseModelSummaries, stripSummaryBlock } from './parse'\n\nexport type ModelResolution = {\n  readonly model: unknown\n  readonly providerId: string\n  readonly modelId: string\n}\n\nexport type ModelResolver = () => Promise<ModelResolution> | ModelResolution\n\nexport type GenerateSource = { sourceType?: string; url?: string; title?: string }\n\nexport type GenerateResult = {\n  readonly text?: string\n  readonly sources?: readonly GenerateSource[]\n}\n\nexport type GenerateFn = (input: {\n  model: unknown\n  tools: Record<string, unknown>\n  prompt: string\n  abortSignal?: AbortSignal\n}) => Promise<GenerateResult>\n\nexport type LoadToolsFn = (providerId: string, maxUses: number) => Promise<Record<string, unknown> | null>\n\n/**\n * `resolveModel`, `generate` and `loadTools` are host-supplied capabilities rather\n * than stored configuration, so they are validated structurally and merged in by\n * the registry at instantiation time. Keeping them in the same schema means the\n * adapter has one source of truth for what it needs, and `readiness()` can say so\n * plainly instead of failing deep inside a search.\n */\nexport const modelOptionsSchema = z.object({\n  maxUses: z.number().int().min(1).max(10).optional(),\n  resolveModel: z.custom<ModelResolver>((value) => typeof value === 'function').optional(),\n  generate: z.custom<GenerateFn>((value) => typeof value === 'function').optional(),\n  loadTools: z.custom<LoadToolsFn>((value) => typeof value === 'function').optional(),\n})\n\nexport type ModelOptions = z.infer<typeof modelOptionsSchema>\n\nconst DEFAULT_MAX_USES = 3\n\n/**\n * A spent quota, a bad key or a suspended account is a configuration problem, not\n * a transient one. Reporting it as `error` tells the scheduler to treat it as\n * bad luck and an operator to wait for it to pass, when the fix is billing or a\n * credential \u2014 so it maps to `unavailable`, which is what \"this source cannot\n * currently work\" means everywhere else in the contract.\n */\nconst NOT_CONFIGURED_MARKERS = [\n  'exceeded your current quota',\n  'insufficient_quota',\n  'billing',\n  'payment required',\n  'invalid api key',\n  'incorrect api key',\n  'invalid_api_key',\n  'authentication',\n  'unauthorized',\n  'account is not active',\n]\n\n/** Shown on the adapter's health row, so it must stay one short clause. */\nconst PROVIDER_CAVEATS: Readonly<Record<string, string>> = {\n  azure: 'searches run via Grounding with Bing, billed separately and outside your geo boundary',\n}\n\nfunction isConfigurationFault(message: string): boolean {\n  const haystack = message.toLowerCase()\n  return NOT_CONFIGURED_MARKERS.some((marker) => haystack.includes(marker))\n}\n\nfunction buildPrompt(request: SearchRequest): string {\n  const constraints: string[] = []\n  if (request.site) constraints.push(`Restrict results to the site ${request.site}.`)\n  if (request.freshness) constraints.push(`Prefer sources published within the last ${request.freshness}.`)\n  if (request.locale) constraints.push(`Prefer sources written for the ${request.locale} locale.`)\n\n  return [\n    `Search the public web for: ${request.query}`,\n    'Use the web_search tool, then answer in two parts.',\n    '',\n    'Part 1: a short prose answer citing the most relevant, authoritative sources.',\n    '',\n    `Part 2: a JSON array in a \\`\\`\\`json code fence with up to ${request.limit} entries, each`,\n    '{\"url\": \"...\", \"title\": \"...\", \"summary\": \"one or two sentences on what this page says about the query\"}.',\n    'Only include URLs you actually retrieved. Write real summaries, not the page title again.',\n    ...constraints,\n  ].join('\\n')\n}\n\nasync function defaultGenerate(input: Parameters<GenerateFn>[0]): Promise<GenerateResult> {\n  const { generateText } = await import('ai')\n  type GenerateTextInput = Parameters<typeof generateText>[0]\n  return (await generateText({\n    model: input.model as GenerateTextInput['model'],\n    tools: input.tools as GenerateTextInput['tools'],\n    prompt: input.prompt,\n    ...(input.abortSignal ? { abortSignal: input.abortSignal } : {}),\n  })) as GenerateResult\n}\n\nexport function createModelAdapter(options: ModelOptions): SearchAdapter {\n  const maxUses = options.maxUses ?? DEFAULT_MAX_USES\n  const resolveModel = options.resolveModel\n  const generate = options.generate ?? defaultGenerate\n  const loadTools = options.loadTools ?? loadNativeSearchTool\n\n  return {\n    id: 'model-native',\n    kind: 'model',\n    capabilities: {\n      search: true,\n      fetch: false,\n      snippets: true,\n      content: false,\n      freshness: true,\n      siteFilter: true,\n      cost: 'llm-tokens',\n    },\n    /** `free` \u2014 resolves the configured model and its tool descriptors; no completion is issued. */\n    probeCost: 'free' as const,\n    readiness: () =>\n      resolveModel ? READY : notReady('no model resolver was provided by the host application'),\n\n    async search(request, context): Promise<SearchOutcome> {\n      if (!resolveModel) {\n        return { status: 'unavailable', reason: 'no model resolver was provided by the host application' }\n      }\n\n      let resolution: ModelResolution\n      try {\n        resolution = await resolveModel()\n      } catch (error) {\n        return {\n          status: 'unavailable',\n          reason: `could not resolve a model: ${error instanceof Error ? error.message : String(error)}`,\n        }\n      }\n\n      const tools = await loadTools(resolution.providerId, maxUses)\n      if (!tools) {\n        return {\n          status: 'unavailable',\n          reason: `provider \"${resolution.providerId}\" has no native web search available`,\n        }\n      }\n\n      let result: GenerateResult\n      try {\n        result = await generate({\n          model: resolution.model,\n          tools,\n          prompt: buildPrompt(request),\n          abortSignal: context.signal,\n        })\n      } catch (error) {\n        if (context.signal.aborted) return { status: 'timeout', reason: 'model search was aborted' }\n        const detail = error instanceof Error ? error.message : String(error)\n        if (isConfigurationFault(detail)) {\n          return { status: 'unavailable', reason: `model search is not usable: ${detail}` }\n        }\n        return { status: 'error', reason: `model search failed: ${detail}`, retriable: true }\n      }\n\n      const text = result.text ?? ''\n      const summaries = new Map(parseModelSummaries(text).map((entry) => [entry.url, entry]))\n\n      // Only URLs the search tool actually returned are trusted. The model's own\n      // JSON is used solely to enrich those, never to introduce a link, because a\n      // fabricated URL that reaches an agent is indistinguishable from a real one.\n      const results: RawResult[] = []\n      const seen = new Set<string>()\n      for (const source of result.sources ?? []) {\n        if (source.sourceType && source.sourceType !== 'url') continue\n        const url = source.url\n        if (!url || seen.has(url)) continue\n        seen.add(url)\n        const enrichment = summaries.get(url)\n        results.push({\n          url,\n          title: enrichment?.title ?? source.title ?? null,\n          snippet: enrichment?.summary ?? null,\n        })\n        if (results.length >= request.limit) break\n      }\n\n      const answer = stripSummaryBlock(text)\n      context.report('ok', `model returned ${results.length} cited source(s)`, {\n        resultCount: results.length,\n      })\n\n      return searchOk(results, answer)\n    },\n\n    async healthCheck() {\n      if (!resolveModel) return { ok: false, detail: 'no model resolver was provided by the host application' }\n      try {\n        const resolution = await resolveModel()\n        const tools = await loadTools(resolution.providerId, maxUses)\n        if (!tools) {\n          return { ok: false, detail: `provider \"${resolution.providerId}\" has no native web search` }\n        }\n        // Azure's native search is Grounding with Bing underneath: it is billed\n        // separately and Microsoft states the Data Protection Addendum does not\n        // cover it, so queries leave the tenant's compliance and geo boundary.\n        // Anyone on Azure for data residency needs to see that, not discover it.\n        const caveat = PROVIDER_CAVEATS[resolution.providerId]\n        return {\n          ok: true,\n          detail: `${resolution.providerId}/${resolution.modelId}${caveat ? ` - ${caveat}` : ''}`,\n        }\n      } catch (error) {\n        return { ok: false, detail: error instanceof Error ? error.message : String(error) }\n      }\n    },\n  }\n}\n\nexport const modelAdapterModule = defineAdapterModule({\n  id: 'model-native',\n  kind: 'model',\n  contractVersion: CONTRACT_VERSION,\n  optionsSchema: modelOptionsSchema,\n  createAdapter: createModelAdapter,\n})\n"],
+  "mappings": "AAAA;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,OAKK;AACP,SAAS,SAAS;AAClB,SAAS,4BAA4B;AACrC,SAAS,qBAAqB,yBAAyB;AAiChD,MAAM,qBAAqB,EAAE,OAAO;AAAA,EACzC,SAAS,EAAE,OAAO,EAAE,IAAI,EAAE,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,SAAS;AAAA,EAClD,cAAc,EAAE,OAAsB,CAAC,UAAU,OAAO,UAAU,UAAU,EAAE,SAAS;AAAA,EACvF,UAAU,EAAE,OAAmB,CAAC,UAAU,OAAO,UAAU,UAAU,EAAE,SAAS;AAAA,EAChF,WAAW,EAAE,OAAoB,CAAC,UAAU,OAAO,UAAU,UAAU,EAAE,SAAS;AACpF,CAAC;AAID,MAAM,mBAAmB;AASzB,MAAM,yBAAyB;AAAA,EAC7B;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AAAA,EACA;AACF;AAGA,MAAM,mBAAqD;AAAA,EACzD,OAAO;AACT;AAEA,SAAS,qBAAqB,SAA0B;AACtD,QAAM,WAAW,QAAQ,YAAY;AACrC,SAAO,uBAAuB,KAAK,CAAC,WAAW,SAAS,SAAS,MAAM,CAAC;AAC1E;AAEA,SAAS,YAAY,SAAgC;AACnD,QAAM,cAAwB,CAAC;AAC/B,MAAI,QAAQ,KAAM,aAAY,KAAK,gCAAgC,QAAQ,IAAI,GAAG;AAClF,MAAI,QAAQ,UAAW,aAAY,KAAK,4CAA4C,QAAQ,SAAS,GAAG;AACxG,MAAI,QAAQ,OAAQ,aAAY,KAAK,kCAAkC,QAAQ,MAAM,UAAU;AAE/F,SAAO;AAAA,IACL,8BAA8B,QAAQ,KAAK;AAAA,IAC3C;AAAA,IACA;AAAA,IACA;AAAA,IACA;AAAA,IACA,8DAA8D,QAAQ,KAAK;AAAA,IAC3E;AAAA,IACA;AAAA,IACA,GAAG;AAAA,EACL,EAAE,KAAK,IAAI;AACb;AAEA,eAAe,gBAAgB,OAA2D;AACxF,QAAM,EAAE,aAAa,IAAI,MAAM,OAAO,IAAI;AAE1C,SAAQ,MAAM,aAAa;AAAA,IACzB,OAAO,MAAM;AAAA,IACb,OAAO,MAAM;AAAA,IACb,QAAQ,MAAM;AAAA,IACd,GAAI,MAAM,cAAc,EAAE,aAAa,MAAM,YAAY,IAAI,CAAC;AAAA,EAChE,CAAC;AACH;AAEO,SAAS,mBAAmB,SAAsC;AACvE,QAAM,UAAU,QAAQ,WAAW;AACnC,QAAM,eAAe,QAAQ;AAC7B,QAAM,WAAW,QAAQ,YAAY;AACrC,QAAM,YAAY,QAAQ,aAAa;AAEvC,SAAO;AAAA,IACL,IAAI;AAAA,IACJ,MAAM;AAAA,IACN,cAAc;AAAA,MACZ,QAAQ;AAAA,MACR,OAAO;AAAA,MACP,UAAU;AAAA,MACV,SAAS;AAAA,MACT,WAAW;AAAA,MACX,YAAY;AAAA,MACZ,MAAM;AAAA,IACR;AAAA;AAAA,IAEA,WAAW;AAAA,IACX,WAAW,MACT,eAAe,QAAQ,SAAS,wDAAwD;AAAA,IAE1F,MAAM,OAAO,SAAS,SAAiC;AACrD,UAAI,CAAC,cAAc;AACjB,eAAO,EAAE,QAAQ,eAAe,QAAQ,yDAAyD;AAAA,MACnG;AAEA,UAAI;AACJ,UAAI;AACF,qBAAa,MAAM,aAAa;AAAA,MAClC,SAAS,OAAO;AACd,eAAO;AAAA,UACL,QAAQ;AAAA,UACR,QAAQ,8BAA8B,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,CAAC;AAAA,QAC9F;AAAA,MACF;AAEA,YAAM,QAAQ,MAAM,UAAU,WAAW,YAAY,OAAO;AAC5D,UAAI,CAAC,OAAO;AACV,eAAO;AAAA,UACL,QAAQ;AAAA,UACR,QAAQ,aAAa,WAAW,UAAU;AAAA,QAC5C;AAAA,MACF;AAEA,UAAI;AACJ,UAAI;AACF,iBAAS,MAAM,SAAS;AAAA,UACtB,OAAO,WAAW;AAAA,UAClB;AAAA,UACA,QAAQ,YAAY,OAAO;AAAA,UAC3B,aAAa,QAAQ;AAAA,QACvB,CAAC;AAAA,MACH,SAAS,OAAO;AACd,YAAI,QAAQ,OAAO,QAAS,QAAO,EAAE,QAAQ,WAAW,QAAQ,2BAA2B;AAC3F,cAAM,SAAS,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK;AACpE,YAAI,qBAAqB,MAAM,GAAG;AAChC,iBAAO,EAAE,QAAQ,eAAe,QAAQ,+BAA+B,MAAM,GAAG;AAAA,QAClF;AACA,eAAO,EAAE,QAAQ,SAAS,QAAQ,wBAAwB,MAAM,IAAI,WAAW,KAAK;AAAA,MACtF;AAEA,YAAM,OAAO,OAAO,QAAQ;AAC5B,YAAM,YAAY,IAAI,IAAI,oBAAoB,IAAI,EAAE,IAAI,CAAC,UAAU,CAAC,MAAM,KAAK,KAAK,CAAC,CAAC;AAKtF,YAAM,UAAuB,CAAC;AAC9B,YAAM,OAAO,oBAAI,IAAY;AAC7B,iBAAW,UAAU,OAAO,WAAW,CAAC,GAAG;AACzC,YAAI,OAAO,cAAc,OAAO,eAAe,MAAO;AACtD,cAAM,MAAM,OAAO;AACnB,YAAI,CAAC,OAAO,KAAK,IAAI,GAAG,EAAG;AAC3B,aAAK,IAAI,GAAG;AACZ,cAAM,aAAa,UAAU,IAAI,GAAG;AACpC,gBAAQ,KAAK;AAAA,UACX;AAAA,UACA,OAAO,YAAY,SAAS,OAAO,SAAS;AAAA,UAC5C,SAAS,YAAY,WAAW;AAAA,QAClC,CAAC;AACD,YAAI,QAAQ,UAAU,QAAQ,MAAO;AAAA,MACvC;AAEA,YAAM,SAAS,kBAAkB,IAAI;AACrC,cAAQ,OAAO,MAAM,kBAAkB,QAAQ,MAAM,oBAAoB;AAAA,QACvE,aAAa,QAAQ;AAAA,MACvB,CAAC;AAED,aAAO,SAAS,SAAS,MAAM;AAAA,IACjC;AAAA,IAEA,MAAM,cAAc;AAClB,UAAI,CAAC,aAAc,QAAO,EAAE,IAAI,OAAO,QAAQ,yDAAyD;AACxG,UAAI;AACF,cAAM,aAAa,MAAM,aAAa;AACtC,cAAM,QAAQ,MAAM,UAAU,WAAW,YAAY,OAAO;AAC5D,YAAI,CAAC,OAAO;AACV,iBAAO,EAAE,IAAI,OAAO,QAAQ,aAAa,WAAW,UAAU,6BAA6B;AAAA,QAC7F;AAKA,cAAM,SAAS,iBAAiB,WAAW,UAAU;AACrD,eAAO;AAAA,UACL,IAAI;AAAA,UACJ,QAAQ,GAAG,WAAW,UAAU,IAAI,WAAW,OAAO,GAAG,SAAS,MAAM,MAAM,KAAK,EAAE;AAAA,QACvF;AAAA,MACF,SAAS,OAAO;AACd,eAAO,EAAE,IAAI,OAAO,QAAQ,iBAAiB,QAAQ,MAAM,UAAU,OAAO,KAAK,EAAE;AAAA,MACrF;AAAA,IACF;AAAA,EACF;AACF;AAEO,MAAM,qBAAqB,oBAAoB;AAAA,EACpD,IAAI;AAAA,EACJ,MAAM;AAAA,EACN,iBAAiB;AAAA,EACjB,eAAe;AAAA,EACf,eAAe;AACjB,CAAC;",
+  "names": []
+}

package/dist/index.js

@@ -0,0 +1,18 @@
+import {
+  createModelAdapter,
+  modelAdapterModule,
+  modelOptionsSchema
+} from "./adapter.js";
+import { parseModelSummaries, stripSummaryBlock } from "./parse.js";
+import { loadNativeSearchTool } from "./nativeTools.js";
+import { modelAdapterModule as modelAdapterModule2 } from "./adapter.js";
+export {
+  createModelAdapter,
+  modelAdapterModule2 as default,
+  loadNativeSearchTool,
+  modelAdapterModule,
+  modelOptionsSchema,
+  parseModelSummaries,
+  stripSummaryBlock
+};
+//# sourceMappingURL=index.js.map

package/dist/index.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/index.ts"],
+  "sourcesContent": ["export {\n  createModelAdapter,\n  modelAdapterModule,\n  modelOptionsSchema,\n  type GenerateFn,\n  type GenerateResult,\n  type GenerateSource,\n  type ModelOptions,\n  type ModelResolution,\n  type ModelResolver,\n} from './adapter'\nexport { parseModelSummaries, stripSummaryBlock, type ModelSummary } from './parse'\nexport { loadNativeSearchTool } from './nativeTools'\n\nexport { modelAdapterModule as default } from './adapter'\n"],
+  "mappings": "AAAA;AAAA,EACE;AAAA,EACA;AAAA,EACA;AAAA,OAOK;AACP,SAAS,qBAAqB,yBAA4C;AAC1E,SAAS,4BAA4B;AAErC,SAA+B,sBAAtBA,2BAAqC;",
+  "names": ["modelAdapterModule"]
+}

package/dist/nativeTools.js

@@ -0,0 +1,31 @@
+async function loadNativeSearchTool(providerId, maxUses) {
+  if (providerId === "anthropic") {
+    try {
+      const { anthropic } = await import("@ai-sdk/anthropic");
+      return { web_search: anthropic.tools.webSearch_20250305({ maxUses }) };
+    } catch {
+      return null;
+    }
+  }
+  if (providerId === "openai") {
+    try {
+      const { openai } = await import("@ai-sdk/openai");
+      return { web_search: openai.tools.webSearch() };
+    } catch {
+      return null;
+    }
+  }
+  if (providerId === "azure") {
+    try {
+      const { azure } = await import("@ai-sdk/azure");
+      return { web_search: azure.tools.webSearch() };
+    } catch {
+      return null;
+    }
+  }
+  return null;
+}
+export {
+  loadNativeSearchTool
+};
+//# sourceMappingURL=nativeTools.js.map

package/dist/nativeTools.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/nativeTools.ts"],
+  "sourcesContent": ["export type NativeToolSet = Record<string, unknown>\n\n/**\n * Loads the provider's own web-search tool lazily.\n *\n * The import is dynamic on purpose: this package is statically imported by the\n * generated adapter registry, and a top-level import of a provider SDK the host\n * did not install would fail the whole registry rather than this one adapter.\n * Both provider packages are optional peers for the same reason.\n */\nexport async function loadNativeSearchTool(\n  providerId: string,\n  maxUses: number,\n): Promise<NativeToolSet | null> {\n  if (providerId === 'anthropic') {\n    try {\n      const { anthropic } = await import('@ai-sdk/anthropic')\n      return { web_search: anthropic.tools.webSearch_20250305({ maxUses }) }\n    } catch {\n      return null\n    }\n  }\n  if (providerId === 'openai') {\n    try {\n      const { openai } = await import('@ai-sdk/openai')\n      return { web_search: openai.tools.webSearch() }\n    } catch {\n      return null\n    }\n  }\n  if (providerId === 'azure') {\n    try {\n      // Azure's web search is a Responses API built-in backed by Grounding with\n      // Bing. `@ai-sdk/azure` calls the Responses API by default, which is what\n      // makes the tool reachable at all \u2014 the OpenAI-compatible Chat Completions\n      // surface Azure used to be wired through cannot carry it.\n      const { azure } = await import('@ai-sdk/azure')\n      return { web_search: azure.tools.webSearch() }\n    } catch {\n      return null\n    }\n  }\n  return null\n}\n"],
+  "mappings": "AAUA,eAAsB,qBACpB,YACA,SAC+B;AAC/B,MAAI,eAAe,aAAa;AAC9B,QAAI;AACF,YAAM,EAAE,UAAU,IAAI,MAAM,OAAO,mBAAmB;AACtD,aAAO,EAAE,YAAY,UAAU,MAAM,mBAAmB,EAAE,QAAQ,CAAC,EAAE;AAAA,IACvE,QAAQ;AACN,aAAO;AAAA,IACT;AAAA,EACF;AACA,MAAI,eAAe,UAAU;AAC3B,QAAI;AACF,YAAM,EAAE,OAAO,IAAI,MAAM,OAAO,gBAAgB;AAChD,aAAO,EAAE,YAAY,OAAO,MAAM,UAAU,EAAE;AAAA,IAChD,QAAQ;AACN,aAAO;AAAA,IACT;AAAA,EACF;AACA,MAAI,eAAe,SAAS;AAC1B,QAAI;AAKF,YAAM,EAAE,MAAM,IAAI,MAAM,OAAO,eAAe;AAC9C,aAAO,EAAE,YAAY,MAAM,MAAM,UAAU,EAAE;AAAA,IAC/C,QAAQ;AACN,aAAO;AAAA,IACT;AAAA,EACF;AACA,SAAO;AACT;",
+  "names": []
+}

package/dist/parse.js

@@ -0,0 +1,47 @@
+const FENCE = /```(?:json)?([\s\S]*?)```/;
+function coerceSummaries(value) {
+  if (!Array.isArray(value)) return [];
+  const summaries = [];
+  for (const item of value) {
+    if (typeof item !== "object" || item === null) continue;
+    const record = item;
+    const url = typeof record.url === "string" ? record.url.trim() : "";
+    if (url.length === 0) continue;
+    summaries.push({
+      url,
+      ...typeof record.title === "string" && record.title.trim().length > 0 ? { title: record.title.trim() } : {},
+      ...typeof record.summary === "string" && record.summary.trim().length > 0 ? { summary: record.summary.trim() } : {}
+    });
+  }
+  return summaries;
+}
+function parseModelSummaries(text) {
+  if (text.trim().length === 0) return [];
+  const fenced = FENCE.exec(text);
+  if (fenced) {
+    try {
+      return coerceSummaries(JSON.parse(fenced[1]));
+    } catch {
+    }
+  }
+  const start = text.indexOf("[");
+  const end = text.lastIndexOf("]");
+  if (start === -1 || end <= start) return [];
+  try {
+    return coerceSummaries(JSON.parse(text.slice(start, end + 1)));
+  } catch {
+    return [];
+  }
+}
+function stripSummaryBlock(text) {
+  const withoutFence = text.replace(FENCE, "").trim();
+  if (withoutFence.length > 0) return withoutFence;
+  const start = text.indexOf("[");
+  if (start <= 0) return "";
+  return text.slice(0, start).trim();
+}
+export {
+  parseModelSummaries,
+  stripSummaryBlock
+};
+//# sourceMappingURL=parse.js.map

package/dist/parse.js.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../src/parse.ts"],
+  "sourcesContent": ["export type ModelSummary = {\n  readonly url: string\n  readonly title?: string\n  readonly summary?: string\n}\n\n/**\n * The leading whitespace stays inside the capture group rather than being eaten\n * by a `\\s*` prefix: `\\s*` and `[\\s\\S]*?` can both match the same whitespace,\n * which makes an unterminated fence backtrack quadratically (CodeQL polynomial\n * ReDoS). The matched span is identical either way, and `JSON.parse` ignores\n * leading whitespace.\n */\nconst FENCE = /```(?:json)?([\\s\\S]*?)```/\n\nfunction coerceSummaries(value: unknown): ModelSummary[] {\n  if (!Array.isArray(value)) return []\n  const summaries: ModelSummary[] = []\n  for (const item of value) {\n    if (typeof item !== 'object' || item === null) continue\n    const record = item as Record<string, unknown>\n    const url = typeof record.url === 'string' ? record.url.trim() : ''\n    if (url.length === 0) continue\n    summaries.push({\n      url,\n      ...(typeof record.title === 'string' && record.title.trim().length > 0\n        ? { title: record.title.trim() }\n        : {}),\n      ...(typeof record.summary === 'string' && record.summary.trim().length > 0\n        ? { summary: record.summary.trim() }\n        : {}),\n    })\n  }\n  return summaries\n}\n\n/**\n * Pulls the per-source summaries out of the model's reply. Best-effort by design:\n * the summaries only ever enrich URLs that the search tool actually returned, so\n * a malformed or absent block costs snippets, never correctness.\n */\nexport function parseModelSummaries(text: string): ModelSummary[] {\n  if (text.trim().length === 0) return []\n\n  const fenced = FENCE.exec(text)\n  if (fenced) {\n    try {\n      return coerceSummaries(JSON.parse(fenced[1]))\n    } catch {\n      // fall through to the bare-array scan\n    }\n  }\n\n  const start = text.indexOf('[')\n  const end = text.lastIndexOf(']')\n  if (start === -1 || end <= start) return []\n  try {\n    return coerceSummaries(JSON.parse(text.slice(start, end + 1)))\n  } catch {\n    return []\n  }\n}\n\n/** Removes the machine-readable block so the prose answer reads naturally. */\nexport function stripSummaryBlock(text: string): string {\n  const withoutFence = text.replace(FENCE, '').trim()\n  if (withoutFence.length > 0) return withoutFence\n  const start = text.indexOf('[')\n  if (start <= 0) return ''\n  return text.slice(0, start).trim()\n}\n"],
+  "mappings": "AAaA,MAAM,QAAQ;AAEd,SAAS,gBAAgB,OAAgC;AACvD,MAAI,CAAC,MAAM,QAAQ,KAAK,EAAG,QAAO,CAAC;AACnC,QAAM,YAA4B,CAAC;AACnC,aAAW,QAAQ,OAAO;AACxB,QAAI,OAAO,SAAS,YAAY,SAAS,KAAM;AAC/C,UAAM,SAAS;AACf,UAAM,MAAM,OAAO,OAAO,QAAQ,WAAW,OAAO,IAAI,KAAK,IAAI;AACjE,QAAI,IAAI,WAAW,EAAG;AACtB,cAAU,KAAK;AAAA,MACb;AAAA,MACA,GAAI,OAAO,OAAO,UAAU,YAAY,OAAO,MAAM,KAAK,EAAE,SAAS,IACjE,EAAE,OAAO,OAAO,MAAM,KAAK,EAAE,IAC7B,CAAC;AAAA,MACL,GAAI,OAAO,OAAO,YAAY,YAAY,OAAO,QAAQ,KAAK,EAAE,SAAS,IACrE,EAAE,SAAS,OAAO,QAAQ,KAAK,EAAE,IACjC,CAAC;AAAA,IACP,CAAC;AAAA,EACH;AACA,SAAO;AACT;AAOO,SAAS,oBAAoB,MAA8B;AAChE,MAAI,KAAK,KAAK,EAAE,WAAW,EAAG,QAAO,CAAC;AAEtC,QAAM,SAAS,MAAM,KAAK,IAAI;AAC9B,MAAI,QAAQ;AACV,QAAI;AACF,aAAO,gBAAgB,KAAK,MAAM,OAAO,CAAC,CAAC,CAAC;AAAA,IAC9C,QAAQ;AAAA,IAER;AAAA,EACF;AAEA,QAAM,QAAQ,KAAK,QAAQ,GAAG;AAC9B,QAAM,MAAM,KAAK,YAAY,GAAG;AAChC,MAAI,UAAU,MAAM,OAAO,MAAO,QAAO,CAAC;AAC1C,MAAI;AACF,WAAO,gBAAgB,KAAK,MAAM,KAAK,MAAM,OAAO,MAAM,CAAC,CAAC,CAAC;AAAA,EAC/D,QAAQ;AACN,WAAO,CAAC;AAAA,EACV;AACF;AAGO,SAAS,kBAAkB,MAAsB;AACtD,QAAM,eAAe,KAAK,QAAQ,OAAO,EAAE,EAAE,KAAK;AAClD,MAAI,aAAa,SAAS,EAAG,QAAO;AACpC,QAAM,QAAQ,KAAK,QAAQ,GAAG;AAC9B,MAAI,SAAS,EAAG,QAAO;AACvB,SAAO,KAAK,MAAM,GAAG,KAAK,EAAE,KAAK;AACnC;",
+  "names": []
+}

package/jest.config.cjs

@@ -0,0 +1,28 @@
+/** @type {import('jest').Config} */
+const base = require('../../jest.config.base.cjs')
+
+module.exports = {
+  ...base,
+  preset: 'ts-jest',
+  testEnvironment: 'node',
+  watchman: false,
+  rootDir: '.',
+  moduleFileExtensions: ['ts', 'tsx', 'js', 'jsx', 'json'],
+  moduleNameMapper: {
+    '^@open-mercato/web-research/(.*)$': '<rootDir>/../web-research/src/$1',
+    '^@open-mercato/web-research$': '<rootDir>/../web-research/src/index.ts',
+  },
+  transform: {
+    '^.+\\.(t|j)sx?$': [
+      'ts-jest',
+      {
+        tsconfig: {
+          rootDir: '.',
+          ignoreDeprecations: '6.0',
+        },
+      },
+    ],
+  },
+  testMatch: ['<rootDir>/src/**/__tests__/**/*.test.(ts|tsx)'],
+  passWithNoTests: true,
+}

package/package.json

@@ -0,0 +1,78 @@
+{
+  "name": "@open-mercato/web-research-model",
+  "version": "0.7.1-develop.7194.1.ab4fc81f82",
+  "license": "MIT",
+  "type": "module",
+  "main": "./dist/index.js",
+  "openMercato": {
+    "webResearchAdapter": {
+      "id": "model-native"
+    }
+  },
+  "scripts": {
+    "build": "node build.mjs",
+    "watch": "node watch.mjs",
+    "test": "jest --config jest.config.cjs",
+    "typecheck": "tsc --noEmit"
+  },
+  "exports": {
+    ".": {
+      "types": "./src/index.ts",
+      "default": "./dist/index.js"
+    },
+    "./*.ts": {
+      "types": "./src/*.ts",
+      "default": "./dist/*.js"
+    },
+    "./*": {
+      "types": [
+        "./src/*.ts"
+      ],
+      "default": "./dist/*.js"
+    }
+  },
+  "peerDependencies": {
+    "@ai-sdk/anthropic": "^4.0.1",
+    "@ai-sdk/azure": "^4.0.0",
+    "@ai-sdk/openai": "^4.0.2",
+    "@open-mercato/web-research": "^0.6.0",
+    "ai": "^7.0.4"
+  },
+  "peerDependenciesMeta": {
+    "@ai-sdk/anthropic": {
+      "optional": true
+    },
+    "@ai-sdk/azure": {
+      "optional": true
+    },
+    "@ai-sdk/openai": {
+      "optional": true
+    }
+  },
+  "dependencies": {
+    "zod": "^4.4.3"
+  },
+  "devDependencies": {
+    "@ai-sdk/anthropic": "^4.0.1",
+    "@ai-sdk/azure": "^4.0.0",
+    "@ai-sdk/openai": "^4.0.2",
+    "@open-mercato/web-research": "0.7.1-develop.7194.1.ab4fc81f82",
+    "@types/jest": "^30.0.0",
+    "@types/node": "^26.0.1",
+    "ai": "^7.0.4",
+    "esbuild": "^0.28.1",
+    "glob": "^13.0.6",
+    "jest": "^30.4.2",
+    "ts-jest": "^29.4.11",
+    "typescript": "^6.0.3"
+  },
+  "publishConfig": {
+    "access": "public"
+  },
+  "repository": {
+    "type": "git",
+    "url": "https://github.com/open-mercato/open-mercato",
+    "directory": "packages/web-research-model"
+  },
+  "stableVersion": "0.7.0"
+}

package/src/__tests__/model.test.ts

@@ -0,0 +1,248 @@
+import { createTestContext, describeAdapterContract } from '@open-mercato/web-research/testing'
+import { createModelAdapter, modelAdapterModule, type GenerateFn, type ModelResolver } from '../adapter'
+import { parseModelSummaries, stripSummaryBlock } from '../parse'
+
+const anthropicResolver: ModelResolver = () => ({
+  model: { id: 'fake' },
+  providerId: 'anthropic',
+  modelId: 'claude-sonnet-5',
+})
+
+const REPLY = `Open Mercato is a modular commerce platform.
+
+\`\`\`json
+[
+  {"url": "https://a.com/1", "title": "Alpha docs", "summary": "Explains the module system in detail."},
+  {"url": "https://b.com/2", "title": "Beta post", "summary": "Covers pricing."}
+]
+\`\`\``
+
+const generate =
+  (result: Awaited<ReturnType<GenerateFn>>): GenerateFn =>
+  async () =>
+    result
+
+/** Stands in for the provider SDK so these tests never load one. */
+const loadTools = async () => ({ web_search: {} })
+
+describe('parseModelSummaries', () => {
+  it('reads a fenced JSON block', () => {
+    expect(parseModelSummaries(REPLY)).toEqual([
+      { url: 'https://a.com/1', title: 'Alpha docs', summary: 'Explains the module system in detail.' },
+      { url: 'https://b.com/2', title: 'Beta post', summary: 'Covers pricing.' },
+    ])
+  })
+
+  it('reads a bare array when the model skipped the fence', () => {
+    expect(parseModelSummaries('Here you go: [{"url":"https://x/1","summary":"s"}]')).toEqual([
+      { url: 'https://x/1', summary: 's' },
+    ])
+  })
+
+  it('returns nothing for malformed or absent JSON', () => {
+    expect(parseModelSummaries('no json here')).toEqual([])
+    expect(parseModelSummaries('```json\n{ broken ]\n```')).toEqual([])
+  })
+
+  it('drops entries without a url', () => {
+    expect(parseModelSummaries('[{"summary":"orphan"},{"url":"https://x/1"}]')).toEqual([
+      { url: 'https://x/1' },
+    ])
+  })
+})
+
+describe('stripSummaryBlock', () => {
+  it('keeps the prose and drops the machine-readable block', () => {
+    expect(stripSummaryBlock(REPLY)).toBe('Open Mercato is a modular commerce platform.')
+  })
+})
+
+describe('model adapter', () => {
+  it('is unavailable when the host provided no resolver', async () => {
+    const adapter = createModelAdapter({})
+    expect(adapter.readiness().ready).toBe(false)
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'unavailable',
+    })
+  })
+
+  it('attaches model summaries to the URLs the search tool returned', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({
+        text: REPLY,
+        sources: [
+          { sourceType: 'url', url: 'https://a.com/1', title: 'Raw title' },
+          { sourceType: 'url', url: 'https://b.com/2' },
+        ],
+      }),
+    })
+
+    const outcome = await adapter.search({ query: 'open mercato', limit: 5 }, createTestContext())
+
+    expect(outcome.status).toBe('ok')
+    if (outcome.status !== 'ok') return
+    // The old implementation returned snippet: '' here, which is the whole reason
+    // the agent had to spend a web_fetch per link to learn anything.
+    expect(outcome.results).toEqual([
+      { url: 'https://a.com/1', title: 'Alpha docs', snippet: 'Explains the module system in detail.' },
+      { url: 'https://b.com/2', title: 'Beta post', snippet: 'Covers pricing.' },
+    ])
+    expect(outcome.answer).toBe('Open Mercato is a modular commerce platform.')
+  })
+
+  it('never introduces a URL the search tool did not return', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({
+        text: '[{"url":"https://hallucinated.example/x","summary":"invented"}]',
+        sources: [{ sourceType: 'url', url: 'https://real.com/1' }],
+      }),
+    })
+
+    const outcome = await adapter.search({ query: 'x', limit: 5 }, createTestContext())
+
+    expect(outcome.status).toBe('ok')
+    if (outcome.status !== 'ok') return
+    expect(outcome.results.map((result) => result.url)).toEqual(['https://real.com/1'])
+  })
+
+  it('keeps a prose answer that cited nothing', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({ text: 'A plain answer with no citations.', sources: [] }),
+    })
+
+    const outcome = await adapter.search({ query: 'x', limit: 5 }, createTestContext())
+
+    expect(outcome).toMatchObject({ status: 'ok', results: [], answer: 'A plain answer with no citations.' })
+  })
+
+  it('is empty when the model produced neither links nor prose', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({ text: '', sources: [] }),
+    })
+
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'empty',
+    })
+  })
+
+  it('reports a provider without native search as unavailable, not an error', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: () => ({ model: {}, providerId: 'ollama', modelId: 'llama' }),
+      loadTools: async () => null,
+      generate: generate({ text: 'x', sources: [] }),
+    })
+
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'unavailable',
+    })
+  })
+
+  it('surfaces a resolver failure as unavailable', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: () => {
+        throw new Error('no API key in this process')
+      },
+    })
+
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'unavailable',
+      reason: expect.stringContaining('no API key in this process'),
+    })
+  })
+
+  it('reports a spent quota as unavailable rather than a retriable error', async () => {
+    // Seen live against OpenAI: reported as `error` with retriable: true, which
+    // reads as bad luck when the fix is billing.
+    const adapter = createModelAdapter({
+      resolveModel: () => ({ model: {}, providerId: 'openai', modelId: 'gpt' }),
+      loadTools: async () => ({ web_search: {} }),
+      generate: async () => {
+        throw new Error(
+          'Failed after 3 attempts. Last error: AI_APICallError: You exceeded your current quota, please check your plan and billing details.',
+        )
+      },
+    })
+
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'unavailable',
+    })
+  })
+
+  it('still reports a genuine transient failure as a retriable error', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: () => ({ model: {}, providerId: 'openai', modelId: 'gpt' }),
+      loadTools: async () => ({ web_search: {} }),
+      generate: async () => {
+        throw new Error('socket hang up')
+      },
+    })
+
+    await expect(adapter.search({ query: 'x', limit: 5 }, createTestContext())).resolves.toMatchObject({
+      status: 'error',
+      retriable: true,
+    })
+  })
+
+  it('passes the abort signal down and reports a timeout when aborted', async () => {
+    const controller = new AbortController()
+    let received: AbortSignal | undefined
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: async (input) => {
+        received = input.abortSignal
+        controller.abort()
+        throw new Error('aborted by test')
+      },
+    })
+
+    const outcome = await adapter.search(
+      { query: 'x', limit: 5 },
+      createTestContext({ signal: controller.signal }),
+    )
+
+    expect(received).toBe(controller.signal)
+    expect(outcome.status).toBe('timeout')
+  })
+
+  it('caps results at the requested limit', async () => {
+    const adapter = createModelAdapter({
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({
+        text: '',
+        sources: Array.from({ length: 10 }, (_, index) => ({
+          sourceType: 'url',
+          url: `https://e.com/${index}`,
+        })),
+      }),
+    })
+
+    const outcome = await adapter.search({ query: 'x', limit: 3 }, createTestContext())
+
+    expect(outcome.status).toBe('ok')
+    if (outcome.status !== 'ok') return
+    expect(outcome.results).toHaveLength(3)
+  })
+})
+
+describeAdapterContract(
+  modelAdapterModule,
+  {
+    unconfiguredOptions: {},
+    configuredOptions: {
+      resolveModel: anthropicResolver,
+      loadTools,
+      generate: generate({ text: '', sources: [{ sourceType: 'url', url: 'https://e.com/1' }] }),
+    },
+  },
+  { describe, it, expect: expect as never },
+)

package/src/__tests__/nativeTools.test.ts

@@ -0,0 +1,52 @@
+/**
+ * Covers the provider→tool mapping only.
+ *
+ * The SDK packages are ESM-only and this suite runs through CJS ts-jest, so a
+ * real `await import()` degrades to `require()` and lands in the catch — which
+ * is exactly why every provider silently looked like it had no native search
+ * here. The modules are mocked so the branch logic is asserted; that the real
+ * import resolves is proved at runtime, not by this file.
+ */
+
+jest.mock(
+  '@ai-sdk/azure',
+  () => ({ azure: { tools: { webSearch: () => ({ kind: 'azure-web-search' }) } } }),
+  { virtual: true },
+)
+jest.mock(
+  '@ai-sdk/openai',
+  () => ({ openai: { tools: { webSearch: () => ({ kind: 'openai-web-search' }) } } }),
+  { virtual: true },
+)
+jest.mock(
+  '@ai-sdk/anthropic',
+  () => ({
+    anthropic: { tools: { webSearch_20250305: () => ({ kind: 'anthropic-web-search' }) } },
+  }),
+  { virtual: true },
+)
+
+import { loadNativeSearchTool } from '../nativeTools'
+
+describe('loadNativeSearchTool', () => {
+  it('offers native search on Azure', async () => {
+    // Azure has had a Responses API `web_search` built-in for a while, but this
+    // returned null for it, so an agent on Foundry was told its provider had no
+    // web search at all no matter how capable the deployment was.
+    const tools = await loadNativeSearchTool('azure', 3)
+    expect(tools).not.toBeNull()
+    expect(tools?.web_search).toEqual({ kind: 'azure-web-search' })
+  })
+
+  it('offers native search on OpenAI and Anthropic', async () => {
+    expect((await loadNativeSearchTool('openai', 3))?.web_search).toEqual({ kind: 'openai-web-search' })
+    expect((await loadNativeSearchTool('anthropic', 3))?.web_search).toEqual({
+      kind: 'anthropic-web-search',
+    })
+  })
+
+  it('returns null for a provider that has none', async () => {
+    expect(await loadNativeSearchTool('ollama', 3)).toBeNull()
+    expect(await loadNativeSearchTool('lm-studio', 3)).toBeNull()
+  })
+})

package/src/adapter.ts

@@ -0,0 +1,241 @@
+import {
+  CONTRACT_VERSION,
+  READY,
+  defineAdapterModule,
+  notReady,
+  searchOk,
+  type RawResult,
+  type SearchAdapter,
+  type SearchOutcome,
+  type SearchRequest,
+} from '@open-mercato/web-research'
+import { z } from 'zod'
+import { loadNativeSearchTool } from './nativeTools'
+import { parseModelSummaries, stripSummaryBlock } from './parse'
+
+export type ModelResolution = {
+  readonly model: unknown
+  readonly providerId: string
+  readonly modelId: string
+}
+
+export type ModelResolver = () => Promise<ModelResolution> | ModelResolution
+
+export type GenerateSource = { sourceType?: string; url?: string; title?: string }
+
+export type GenerateResult = {
+  readonly text?: string
+  readonly sources?: readonly GenerateSource[]
+}
+
+export type GenerateFn = (input: {
+  model: unknown
+  tools: Record<string, unknown>
+  prompt: string
+  abortSignal?: AbortSignal
+}) => Promise<GenerateResult>
+
+export type LoadToolsFn = (providerId: string, maxUses: number) => Promise<Record<string, unknown> | null>
+
+/**
+ * `resolveModel`, `generate` and `loadTools` are host-supplied capabilities rather
+ * than stored configuration, so they are validated structurally and merged in by
+ * the registry at instantiation time. Keeping them in the same schema means the
+ * adapter has one source of truth for what it needs, and `readiness()` can say so
+ * plainly instead of failing deep inside a search.
+ */
+export const modelOptionsSchema = z.object({
+  maxUses: z.number().int().min(1).max(10).optional(),
+  resolveModel: z.custom<ModelResolver>((value) => typeof value === 'function').optional(),
+  generate: z.custom<GenerateFn>((value) => typeof value === 'function').optional(),
+  loadTools: z.custom<LoadToolsFn>((value) => typeof value === 'function').optional(),
+})
+
+export type ModelOptions = z.infer<typeof modelOptionsSchema>
+
+const DEFAULT_MAX_USES = 3
+
+/**
+ * A spent quota, a bad key or a suspended account is a configuration problem, not
+ * a transient one. Reporting it as `error` tells the scheduler to treat it as
+ * bad luck and an operator to wait for it to pass, when the fix is billing or a
+ * credential — so it maps to `unavailable`, which is what "this source cannot
+ * currently work" means everywhere else in the contract.
+ */
+const NOT_CONFIGURED_MARKERS = [
+  'exceeded your current quota',
+  'insufficient_quota',
+  'billing',
+  'payment required',
+  'invalid api key',
+  'incorrect api key',
+  'invalid_api_key',
+  'authentication',
+  'unauthorized',
+  'account is not active',
+]
+
+/** Shown on the adapter's health row, so it must stay one short clause. */
+const PROVIDER_CAVEATS: Readonly<Record<string, string>> = {
+  azure: 'searches run via Grounding with Bing, billed separately and outside your geo boundary',
+}
+
+function isConfigurationFault(message: string): boolean {
+  const haystack = message.toLowerCase()
+  return NOT_CONFIGURED_MARKERS.some((marker) => haystack.includes(marker))
+}
+
+function buildPrompt(request: SearchRequest): string {
+  const constraints: string[] = []
+  if (request.site) constraints.push(`Restrict results to the site ${request.site}.`)
+  if (request.freshness) constraints.push(`Prefer sources published within the last ${request.freshness}.`)
+  if (request.locale) constraints.push(`Prefer sources written for the ${request.locale} locale.`)
+
+  return [
+    `Search the public web for: ${request.query}`,
+    'Use the web_search tool, then answer in two parts.',
+    '',
+    'Part 1: a short prose answer citing the most relevant, authoritative sources.',
+    '',
+    `Part 2: a JSON array in a \`\`\`json code fence with up to ${request.limit} entries, each`,
+    '{"url": "...", "title": "...", "summary": "one or two sentences on what this page says about the query"}.',
+    'Only include URLs you actually retrieved. Write real summaries, not the page title again.',
+    ...constraints,
+  ].join('\n')
+}
+
+async function defaultGenerate(input: Parameters<GenerateFn>[0]): Promise<GenerateResult> {
+  const { generateText } = await import('ai')
+  type GenerateTextInput = Parameters<typeof generateText>[0]
+  return (await generateText({
+    model: input.model as GenerateTextInput['model'],
+    tools: input.tools as GenerateTextInput['tools'],
+    prompt: input.prompt,
+    ...(input.abortSignal ? { abortSignal: input.abortSignal } : {}),
+  })) as GenerateResult
+}
+
+export function createModelAdapter(options: ModelOptions): SearchAdapter {
+  const maxUses = options.maxUses ?? DEFAULT_MAX_USES
+  const resolveModel = options.resolveModel
+  const generate = options.generate ?? defaultGenerate
+  const loadTools = options.loadTools ?? loadNativeSearchTool
+
+  return {
+    id: 'model-native',
+    kind: 'model',
+    capabilities: {
+      search: true,
+      fetch: false,
+      snippets: true,
+      content: false,
+      freshness: true,
+      siteFilter: true,
+      cost: 'llm-tokens',
+    },
+    /** `free` — resolves the configured model and its tool descriptors; no completion is issued. */
+    probeCost: 'free' as const,
+    readiness: () =>
+      resolveModel ? READY : notReady('no model resolver was provided by the host application'),
+
+    async search(request, context): Promise<SearchOutcome> {
+      if (!resolveModel) {
+        return { status: 'unavailable', reason: 'no model resolver was provided by the host application' }
+      }
+
+      let resolution: ModelResolution
+      try {
+        resolution = await resolveModel()
+      } catch (error) {
+        return {
+          status: 'unavailable',
+          reason: `could not resolve a model: ${error instanceof Error ? error.message : String(error)}`,
+        }
+      }
+
+      const tools = await loadTools(resolution.providerId, maxUses)
+      if (!tools) {
+        return {
+          status: 'unavailable',
+          reason: `provider "${resolution.providerId}" has no native web search available`,
+        }
+      }
+
+      let result: GenerateResult
+      try {
+        result = await generate({
+          model: resolution.model,
+          tools,
+          prompt: buildPrompt(request),
+          abortSignal: context.signal,
+        })
+      } catch (error) {
+        if (context.signal.aborted) return { status: 'timeout', reason: 'model search was aborted' }
+        const detail = error instanceof Error ? error.message : String(error)
+        if (isConfigurationFault(detail)) {
+          return { status: 'unavailable', reason: `model search is not usable: ${detail}` }
+        }
+        return { status: 'error', reason: `model search failed: ${detail}`, retriable: true }
+      }
+
+      const text = result.text ?? ''
+      const summaries = new Map(parseModelSummaries(text).map((entry) => [entry.url, entry]))
+
+      // Only URLs the search tool actually returned are trusted. The model's own
+      // JSON is used solely to enrich those, never to introduce a link, because a
+      // fabricated URL that reaches an agent is indistinguishable from a real one.
+      const results: RawResult[] = []
+      const seen = new Set<string>()
+      for (const source of result.sources ?? []) {
+        if (source.sourceType && source.sourceType !== 'url') continue
+        const url = source.url
+        if (!url || seen.has(url)) continue
+        seen.add(url)
+        const enrichment = summaries.get(url)
+        results.push({
+          url,
+          title: enrichment?.title ?? source.title ?? null,
+          snippet: enrichment?.summary ?? null,
+        })
+        if (results.length >= request.limit) break
+      }
+
+      const answer = stripSummaryBlock(text)
+      context.report('ok', `model returned ${results.length} cited source(s)`, {
+        resultCount: results.length,
+      })
+
+      return searchOk(results, answer)
+    },
+
+    async healthCheck() {
+      if (!resolveModel) return { ok: false, detail: 'no model resolver was provided by the host application' }
+      try {
+        const resolution = await resolveModel()
+        const tools = await loadTools(resolution.providerId, maxUses)
+        if (!tools) {
+          return { ok: false, detail: `provider "${resolution.providerId}" has no native web search` }
+        }
+        // Azure's native search is Grounding with Bing underneath: it is billed
+        // separately and Microsoft states the Data Protection Addendum does not
+        // cover it, so queries leave the tenant's compliance and geo boundary.
+        // Anyone on Azure for data residency needs to see that, not discover it.
+        const caveat = PROVIDER_CAVEATS[resolution.providerId]
+        return {
+          ok: true,
+          detail: `${resolution.providerId}/${resolution.modelId}${caveat ? ` - ${caveat}` : ''}`,
+        }
+      } catch (error) {
+        return { ok: false, detail: error instanceof Error ? error.message : String(error) }
+      }
+    },
+  }
+}
+
+export const modelAdapterModule = defineAdapterModule({
+  id: 'model-native',
+  kind: 'model',
+  contractVersion: CONTRACT_VERSION,
+  optionsSchema: modelOptionsSchema,
+  createAdapter: createModelAdapter,
+})

package/src/index.ts

@@ -0,0 +1,15 @@
+export {
+  createModelAdapter,
+  modelAdapterModule,
+  modelOptionsSchema,
+  type GenerateFn,
+  type GenerateResult,
+  type GenerateSource,
+  type ModelOptions,
+  type ModelResolution,
+  type ModelResolver,
+} from './adapter'
+export { parseModelSummaries, stripSummaryBlock, type ModelSummary } from './parse'
+export { loadNativeSearchTool } from './nativeTools'
+
+export { modelAdapterModule as default } from './adapter'

package/src/nativeTools.ts

@@ -0,0 +1,44 @@
+export type NativeToolSet = Record<string, unknown>
+
+/**
+ * Loads the provider's own web-search tool lazily.
+ *
+ * The import is dynamic on purpose: this package is statically imported by the
+ * generated adapter registry, and a top-level import of a provider SDK the host
+ * did not install would fail the whole registry rather than this one adapter.
+ * Both provider packages are optional peers for the same reason.
+ */
+export async function loadNativeSearchTool(
+  providerId: string,
+  maxUses: number,
+): Promise<NativeToolSet | null> {
+  if (providerId === 'anthropic') {
+    try {
+      const { anthropic } = await import('@ai-sdk/anthropic')
+      return { web_search: anthropic.tools.webSearch_20250305({ maxUses }) }
+    } catch {
+      return null
+    }
+  }
+  if (providerId === 'openai') {
+    try {
+      const { openai } = await import('@ai-sdk/openai')
+      return { web_search: openai.tools.webSearch() }
+    } catch {
+      return null
+    }
+  }
+  if (providerId === 'azure') {
+    try {
+      // Azure's web search is a Responses API built-in backed by Grounding with
+      // Bing. `@ai-sdk/azure` calls the Responses API by default, which is what
+      // makes the tool reachable at all — the OpenAI-compatible Chat Completions
+      // surface Azure used to be wired through cannot carry it.
+      const { azure } = await import('@ai-sdk/azure')
+      return { web_search: azure.tools.webSearch() }
+    } catch {
+      return null
+    }
+  }
+  return null
+}

package/src/parse.ts

@@ -0,0 +1,71 @@
+export type ModelSummary = {
+  readonly url: string
+  readonly title?: string
+  readonly summary?: string
+}
+
+/**
+ * The leading whitespace stays inside the capture group rather than being eaten
+ * by a `\s*` prefix: `\s*` and `[\s\S]*?` can both match the same whitespace,
+ * which makes an unterminated fence backtrack quadratically (CodeQL polynomial
+ * ReDoS). The matched span is identical either way, and `JSON.parse` ignores
+ * leading whitespace.
+ */
+const FENCE = /```(?:json)?([\s\S]*?)```/
+
+function coerceSummaries(value: unknown): ModelSummary[] {
+  if (!Array.isArray(value)) return []
+  const summaries: ModelSummary[] = []
+  for (const item of value) {
+    if (typeof item !== 'object' || item === null) continue
+    const record = item as Record<string, unknown>
+    const url = typeof record.url === 'string' ? record.url.trim() : ''
+    if (url.length === 0) continue
+    summaries.push({
+      url,
+      ...(typeof record.title === 'string' && record.title.trim().length > 0
+        ? { title: record.title.trim() }
+        : {}),
+      ...(typeof record.summary === 'string' && record.summary.trim().length > 0
+        ? { summary: record.summary.trim() }
+        : {}),
+    })
+  }
+  return summaries
+}
+
+/**
+ * Pulls the per-source summaries out of the model's reply. Best-effort by design:
+ * the summaries only ever enrich URLs that the search tool actually returned, so
+ * a malformed or absent block costs snippets, never correctness.
+ */
+export function parseModelSummaries(text: string): ModelSummary[] {
+  if (text.trim().length === 0) return []
+
+  const fenced = FENCE.exec(text)
+  if (fenced) {
+    try {
+      return coerceSummaries(JSON.parse(fenced[1]))
+    } catch {
+      // fall through to the bare-array scan
+    }
+  }
+
+  const start = text.indexOf('[')
+  const end = text.lastIndexOf(']')
+  if (start === -1 || end <= start) return []
+  try {
+    return coerceSummaries(JSON.parse(text.slice(start, end + 1)))
+  } catch {
+    return []
+  }
+}
+
+/** Removes the machine-readable block so the prose answer reads naturally. */
+export function stripSummaryBlock(text: string): string {
+  const withoutFence = text.replace(FENCE, '').trim()
+  if (withoutFence.length > 0) return withoutFence
+  const start = text.indexOf('[')
+  if (start <= 0) return ''
+  return text.slice(0, start).trim()
+}

package/tsconfig.json

@@ -0,0 +1,10 @@
+{
+  "$schema": "https://json.schemastore.org/tsconfig",
+  "extends": "../../tsconfig.base.json",
+  "compilerOptions": {
+    "noEmit": true,
+    "types": ["node"]
+  },
+  "include": ["src/**/*"],
+  "exclude": ["node_modules", "dist", "**/__tests__/**"]
+}

package/watch.mjs

@@ -0,0 +1,7 @@
+import { dirname } from 'node:path'
+import { fileURLToPath } from 'node:url'
+import { watch } from '../../scripts/watch.mjs'
+
+const __dirname = dirname(fileURLToPath(import.meta.url))
+
+watch(__dirname)