MoonshotAI · bj456736 · Jun 25, 2026 · Jun 25, 2026 · chatgpt-codex-connector · Jun 25, 2026
diff --git a/packages/agent-core/src/tools/builtin/web/fetch-url.md b/packages/agent-core/src/tools/builtin/web/fetch-url.md
@@ -1,3 +1,3 @@
-Fetch content from a URL. Returns the main text content extracted from the page. Use this when you need to read a specific web page.
+Fetch content from a URL. Returns the main text content extracted from the page, or the image data if the URL points to an image file. Use this when you need to read a specific web page or image.
 
 Only public `http`/`https` URLs are supported. Requests to private, loopback, or link-local addresses are refused, and responses larger than 10 MiB are rejected.
diff --git a/packages/agent-core/src/tools/builtin/web/fetch-url.ts b/packages/agent-core/src/tools/builtin/web/fetch-url.ts
@@ -10,7 +10,7 @@
 
 import type { BuiltinTool } from '../../../agent/tool';
 import { ToolAccesses } from '../../../loop/tool-access';
-import type { ExecutableToolContext, ExecutableToolResult, ToolExecution } from '../../../loop/types';
+import type { ContentPart, ExecutableToolContext, ExecutableToolResult, ToolExecution } from '../../../loop/types';
 import { toInputJsonSchema } from '../../support/input-schema';
 import { literalRulePattern, matchesGlobRuleSubject } from '../../support/rule-match';
 import { ToolResultBuilder } from '../../support/result-builder';
@@ -25,14 +25,29 @@
  *   returned verbatim, in full.
  * - `extracted` — the body was an HTML page; only the main article text
  *   was extracted and returned.
+ * - `image` — the body is an image file; the binary data is returned
+ *   as base64-encoded content for multimodal model input.
  */
-export type UrlFetchKind = 'passthrough' | 'extracted';
+export type UrlFetchKind = 'passthrough' | 'extracted' | 'image';
+
+export interface PageMetadata {
+  /** The URL that was fetched. */
+  url: string;
+  /** The title of the page, if available. */
+  title?: string;
+  /** The MIME type of the response. */
+  mime?: string;
+}
 
 export interface UrlFetchResult {
-  /** The text handed to the LLM. */
+  /** The text handed to the LLM, or empty string for image content. */
   content: string;
-  /** Whether `content` is a verbatim passthrough or extracted main text. */
+  /** Whether content is a verbatim passthrough, extracted main text, or image data. */
   kind: UrlFetchKind;
+  /** Image data as base64, when kind is 'image'. */
+  image?: { mimeType: string; base64: string };
+  /** Page metadata for the fetched resource, aligning with internal API shape. */
+  page?: PageMetadata;
 }
 
 export interface UrlFetcher {
@@ -84,12 +99,20 @@
 
   private async execution(
     args: FetchURLInput,
-    {
-    toolCallId,
-    }: ExecutableToolContext,
+    { toolCallId }: ExecutableToolContext,
   ): Promise<ExecutableToolResult> {
     try {
-      const { content, kind } = await this.fetcher.fetch(args.url, { toolCallId });
+      const { content, kind, image } = await this.fetcher.fetch(args.url, { toolCallId });
+
+      if (image) {
+        const output: ContentPart[] = [
+          {
+            type: 'image_url',
+            imageUrl: { url: `data:${image.mimeType};base64,${image.base64}`, id: args.url },
+          },
+        ];
+        return { output, isError: false };
+      }
 
       if (!content) {
         return {

diff --git a/packages/agent-core/src/tools/providers/local-fetch-url.ts b/packages/agent-core/src/tools/providers/local-fetch-url.ts
@@ -7,7 +7,8 @@
  *   3. Reject responses larger than `maxBytes` (content-length first,
  *      then measured body length as a defensive second check).
  *   4. `text/plain` / `text/markdown` → passthrough verbatim.
- *   5. Otherwise (assumed HTML) → run Readability over a linkedom
+ *   5. `image/*` → download binary, encode as base64, return as image kind.
+ *   6. Otherwise (assumed HTML) → run Readability over a linkedom
  *      document. Return `# ${title}\n\n${text}` (title omitted when
  *      absent). If extraction yields no meaningful text, fall back to
  *      common content containers (`<article>` / `<main>` / `<body>`)
@@ -172,6 +173,26 @@ export class LocalFetchURLProvider implements UrlFetcher {
       }
     }
 
+    const contentType = (response.headers.get('content-type') ?? '').toLowerCase();
+
+    // Handle image content types
+    if (contentType.startsWith('image/')) {
+      const arrayBuffer = await response.arrayBuffer();
+      const actualBytes = arrayBuffer.byteLength;
+      if (actualBytes > this.maxBytes) {
+        throw new Error(
+          `Response body too large: ${String(actualBytes)} bytes exceeds maxBytes (${String(this.maxBytes)}).`,
+        );
+      }
+      const base64 = Buffer.from(arrayBuffer).toString('base64');
+      return {
+        content: '',
+        kind: 'image',
+        image: { mimeType: contentType, base64 },
+        page: { url, mime: contentType },
+      };
+    }
+
     const body = await response.text();
 
     // Servers may omit content-length — measure again defensively.
@@ -182,12 +203,11 @@ export class LocalFetchURLProvider implements UrlFetcher {
       );
     }
 
-    const contentType = (response.headers.get('content-type') ?? '').toLowerCase();
     if (contentType.startsWith('text/plain') || contentType.startsWith('text/markdown')) {
-      return { content: body, kind: 'passthrough' };
+      return { content: body, kind: 'passthrough', page: { url, mime: contentType } };
     }
 
-    return { content: this.extractMainContent(body), kind: 'extracted' };
+    return { content: this.extractMainContent(body), kind: 'extracted', page: { url, mime: contentType } };
   }
 
   private extractMainContent(html: string): string {

diff --git a/packages/agent-core/test/tools/fetch-url.test.ts b/packages/agent-core/test/tools/fetch-url.test.ts
@@ -12,6 +12,7 @@ import {
   HttpFetchError,
   type UrlFetcher,
 } from '../../src/tools/builtin/web/fetch-url';
+import type { ImageURLPart } from '@moonshot-ai/kosong';
 import { MoonshotFetchURLProvider } from '../../src/tools/providers/moonshot-fetch-url';
 import { toolContentString } from './fixtures/fake-kaos';
 import { executeTool } from './fixtures/execute-tool';
@@ -20,9 +21,11 @@ const signal = new AbortController().signal;
 
 function fakeFetcher(
   content = '',
-  kind: 'passthrough' | 'extracted' = 'extracted',
+  kind: 'passthrough' | 'extracted' | 'image' = 'extracted',
+  image?: { mimeType: string; base64: string },
+  page?: { url: string; mime?: string; title?: string },
 ): UrlFetcher {
-  return { fetch: vi.fn().mockResolvedValue({ content, kind }) };
+  return { fetch: vi.fn().mockResolvedValue({ content, kind, image, page }) };
 }
 
 describe('FetchURLTool', () => {
@@ -119,6 +122,32 @@ describe('FetchURLTool', () => {
     expect((result as { message?: string }).message).toContain('Output is truncated');
   });
 
+  it('returns image_url content part for image kind results', async () => {
+    const fetcher: UrlFetcher = {
+      fetch: vi.fn().mockResolvedValue({
+        content: '',
+        kind: 'image',
+        image: { mimeType: 'image/png', base64: 'base64data' },
+        page: { url: 'https://example.com/image.png', mime: 'image/png' },
+      }),
+    };
+    const tool = new FetchURLTool(fetcher);
+
+    const result = await executeTool(tool, {
+      turnId: 't1',
+      toolCallId: 'c_img',
+      args: { url: 'https://example.com/image.png' },
+      signal,
+    });
+
+    expect(result.isError).toBe(false);
+    const output = (result as { output?: ContentPart[] }).output;
+    expect(output).toHaveLength(1);
+    expect(output?.[0]?.type).toBe('image_url');
+    expect((output?.[0] as ImageURLPart).imageUrl.url).toBe('data:image/png;base64,base64data');
+    expect((output?.[0] as ImageURLPart).imageUrl.id).toBe('https://example.com/image.png');
+  });
+
   it('returns error when fetcher throws', async () => {
     const fetcher: UrlFetcher = {
       fetch: vi.fn().mockRejectedValue(new Error('timeout')),

diff --git a/packages/agent-core/test/tools/providers/local-fetch-url.test.ts b/packages/agent-core/test/tools/providers/local-fetch-url.test.ts
@@ -17,7 +17,30 @@ function htmlResponse(body: string, contentType: string): Response {
   });
 }
 
+function imageResponse(data: Uint8Array, contentType: string): Response {
+  return new Response(data, {
+    status: 200,
+    headers: { 'content-type': contentType },
+  });
+}
+
 describe('LocalFetchURLProvider content kind', () => {
+  it('reports image content as image kind with base64 data', async () => {
+    const imageData = new Uint8Array([0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a]);
+    const fetchImpl = vi
+      .fn<typeof fetch>()
+      .mockResolvedValue(imageResponse(imageData, 'image/png'));
+    const provider = new LocalFetchURLProvider({ fetchImpl });
+
+    const result = await provider.fetch('https://example.com/image.png');
+
+    expect(result.kind).toBe('image');
+    expect(result.image).toBeDefined();
+    expect(result.image?.mimeType).toBe('image/png');
+    expect(result.image?.base64).toBeTruthy();
+    expect(result.page).toEqual({ url: 'https://example.com/image.png', mime: 'image/png' });
+  });
+
   it('reports text/plain bodies as a verbatim passthrough', async () => {
     const fetchImpl = vi
       .fn<typeof fetch>()
@@ -26,7 +49,7 @@ describe('LocalFetchURLProvider content kind', () => {
 
     const result = await provider.fetch('https://example.com/file.txt');
 
-    expect(result).toEqual({ content: 'plain body', kind: 'passthrough' });
+    expect(result).toEqual({ content: 'plain body', kind: 'passthrough', page: { url: 'https://example.com/file.txt', mime: 'text/plain; charset=utf-8' } });
   });
 
   it('reports text/markdown bodies as a verbatim passthrough', async () => {
@@ -37,7 +60,7 @@ describe('LocalFetchURLProvider content kind', () => {
 
     const result = await provider.fetch('https://example.com/readme.md');
 
-    expect(result).toEqual({ content: '# Title\n\nbody', kind: 'passthrough' });
+    expect(result).toEqual({ content: '# Title\n\nbody', kind: 'passthrough', page: { url: 'https://example.com/readme.md', mime: 'text/markdown' } });
   });
 
   it('reports HTML bodies as extracted main content', async () => {

diff --git a/packages/agent-core/test/tools/providers/moonshot-fetch-url.test.ts b/packages/agent-core/test/tools/providers/moonshot-fetch-url.test.ts
@@ -5,9 +5,11 @@ import { MoonshotFetchURLProvider } from '../../../src/tools/providers/moonshot-
 
 function fakeFetcher(
   content = '',
-  kind: 'passthrough' | 'extracted' = 'extracted',
+  kind: 'passthrough' | 'extracted' | 'image' = 'extracted',
+  image?: { mimeType: string; base64: string },
+  page?: { url: string; mime?: string; title?: string },
 ): UrlFetcher {
-  return { fetch: vi.fn().mockResolvedValue({ content, kind }) };
+  return { fetch: vi.fn().mockResolvedValue({ content, kind, image, page }) };
 }
 
 describe('MoonshotFetchURLProvider auth fallback', () => {