chore: webSearch & FetchUrl Sync #1260

.gitignore

@@ -19,6 +19,7 @@ plugins/cdn/
 .worktrees/
 .kimi-code/local.toml
 .kimi-sandbox/
+.vscode/
 
 Dockerfile
 docker-compose.yml

packages/agent-core-v2/src/app/auth/webSearch/providers/moonshot-web-search.ts

@@ -48,18 +48,11 @@ export class MoonshotWebSearchProvider implements WebSearchProvider {
   async search(
     query: string,
     options?: {
-      limit?: number;
-      includeContent?: boolean;
       toolCallId?: string;
       signal?: AbortSignal;
     },
   ): Promise<WebSearchResult[]> {
-    const body = {
-      text_query: query,
-      limit: options?.limit ?? 5,
-      enable_page_crawling: options?.includeContent ?? false,
-      timeout_seconds: 30,
-    };
+    const body = { text_query: query };
     const bodyJson = JSON.stringify(body);
 
     const toolCallId = options?.toolCallId;
@@ -90,7 +83,6 @@ export class MoonshotWebSearchProvider implements WebSearchProvider {
       };
       if (typeof r.date === 'string' && r.date.length > 0) out.date = r.date;
       if (typeof r.site_name === 'string' && r.site_name.length > 0) out.siteName = r.site_name;
-      if (typeof r.content === 'string' && r.content.length > 0) out.content = r.content;
       return out;
     });
   }

packages/agent-core-v2/src/app/auth/webSearch/tools/web-search.ts

@@ -36,15 +36,12 @@ export interface WebSearchResult {
   snippet: string;
   date?: string;
   siteName?: string;
-  content?: string;
 }
 
 export interface WebSearchProvider {
   search(
     query: string,
     options?: {
-      limit?: number;
-      includeContent?: boolean;
       toolCallId?: string;
       signal?: AbortSignal;
     },
@@ -55,23 +52,6 @@ export interface WebSearchProvider {
 
 export const WebSearchInputSchema = z.object({
   query: z.string().describe('The query text to search for.'),
-  limit: z
-    .number()
-    .int()
-    .min(1)
-    .max(20)
-    .default(5)
-    .describe(
-      'The number of results to return. Typically you do not need to set this value. When the results do not contain what you need, you probably want to give a more concrete query.',
-    )
-    .optional(),
-  include_content: z
-    .boolean()
-    .default(false)
-    .describe(
-      'Whether to include the content of the web pages in the results. It can consume a large amount of tokens when this is set to true. You should avoid enabling this when `limit` is set to a large value.',
-    )
-    .optional(),
 });
 
 export type WebSearchInput = z.infer<typeof WebSearchInputSchema>;
@@ -102,18 +82,7 @@ export class WebSearchTool implements BuiltinTool<WebSearchInput> {
     { toolCallId, signal }: ExecutableToolContext,
   ): Promise<ExecutableToolResult> {
     try {
-      const opts: {
-        limit?: number;
-        includeContent?: boolean;
-        toolCallId?: string;
-        signal?: AbortSignal;
-      } = {
-        toolCallId,
-        signal,
-      };
-      if (args.limit !== undefined) opts.limit = args.limit;
-      if (args.include_content !== undefined) opts.includeContent = args.include_content;
-      const results = await this.provider.search(args.query, opts);
+      const results = await this.provider.search(args.query, { toolCallId, signal });
       const builder = new ToolResultBuilder({ maxLineLength: null });
 
       if (results.length === 0) {
@@ -131,7 +100,6 @@ export class WebSearchTool implements BuiltinTool<WebSearchInput> {
         if (result.date) builder.write(`Date: ${result.date}\n`);
         builder.write(`URL: ${result.url}\n`);
         builder.write(`Snippet: ${result.snippet}\n\n`);
-        if (result.content) builder.write(`${result.content}\n\n`);
       }
 
       // Keep the citation reminder next to the data (not just in the static tool

packages/agent-core-v2/src/app/web/tools/fetch-url.ts

@@ -70,15 +70,20 @@ export class FetchURLTool implements BuiltinTool<FetchURLInput> {
       }
 
       const builder = new ToolResultBuilder({ maxLineLength: null });
-      builder.write(content);
-      // Tell the LLM whether it received the whole body or only the
-      // extracted article text, so it can judge how complete the
-      // content is.
-      const message =
+      // Tell the LLM whether it received the whole body or only the extracted
+      // article text, so it can judge how complete the content is, and remind it
+      // to cite this page when it uses the content. Both notes must ride in
+      // `output`: the result's `message` field is dropped from the transcript, so
+      // `output` is the only place the model can read them. Put them at the front
+      // so they survive any downstream truncation of the body.
+      const note =
         kind === 'passthrough'
           ? 'The returned content is the full response body, returned verbatim.'
           : 'The returned content is the main text extracted from the page.';
-      return builder.ok(message);
+      const citeReminder =
+        'If you use it in your answer, cite this page as a markdown link, e.g. [title](url).';
+      builder.write(`${note} ${citeReminder}\n\n${content}`);
+      return builder.ok();
     } catch (error) {
       // An in-flight abort rejects the signal-aware fetch promptly. Re-throw
       // so the executor can classify it (including user cancellation) and

packages/agent-core-v2/test/app/web/fetch-url.test.ts

@@ -78,6 +78,35 @@ describe('FetchURLTool abort signal', () => {
   });
 });
 
+describe('FetchURLTool output note', () => {
+  async function runKind(kind: UrlFetchResult['kind']): Promise<string> {
+    const fetch = vi
+      .fn<UrlFetcher['fetch']>()
+      .mockResolvedValue({ content: 'BODY', kind } satisfies UrlFetchResult);
+    const tool = new FetchURLTool({ fetch });
+    const result = await execute(tool, 'https://example.com', new AbortController().signal);
+    expect(result.isError).toBe(false);
+    if (typeof result.output !== 'string') throw new Error('expected string output');
+    return result.output;
+  }
+
+  it('puts the passthrough note and citation reminder at the front of output', async () => {
+    const output = await runKind('passthrough');
+    expect(output).toBe(
+      'The returned content is the full response body, returned verbatim. ' +
+        'If you use it in your answer, cite this page as a markdown link, e.g. [title](url).\n\nBODY',
+    );
+  });
+
+  it('puts the extracted note and citation reminder at the front of output', async () => {
+    const output = await runKind('extracted');
+    expect(output).toBe(
+      'The returned content is the main text extracted from the page. ' +
+        'If you use it in your answer, cite this page as a markdown link, e.g. [title](url).\n\nBODY',
+    );
+  });
+});
+
 describe('LocalFetchURLProvider abort signal', () => {
   it('passes the signal through to fetchImpl', async () => {
     const controller = new AbortController();

packages/agent-core-v2/test/auth/auth.test.ts

@@ -678,7 +678,7 @@ describe('WebSearchProviderService', () => {
 
     const provider = createService().getWebSearchProvider();
     expect(provider).not.toBeUndefined();
-    const results = await provider!.search('hello', { limit: 2 });
+    const results = await provider!.search('hello');
 
     expect(results).toEqual([
       { title: 'Title', url: 'https://example.com', snippet: 'Snippet' },
@@ -689,6 +689,7 @@ describe('WebSearchProviderService', () => {
     const headers = init.headers as Record<string, string>;
     expect(headers['Authorization']).toBe('Bearer access-token');
     expect(headers['X-Custom']).toBe('yes');
+    expect(JSON.parse(init.body as string)).toEqual({ text_query: 'hello' });
   });
 });