Merge branch 'main' into feature/exa-search

2026-06-15 12:55:51 +00:00 · 2026-04-23 20:04:13 +08:00
parent 93bfdabff1 1173a62301
commit b966eef5a9
284 changed files with 22610 additions and 6199 deletions
--- a/packages/@ant/model-provider/src/shared/tests/openaiConvertMessages.test.ts
+++ b/packages/@ant/model-provider/src/shared/tests/openaiConvertMessages.test.ts
@@ -21,26 +21,22 @@ function makeAssistantMsg(content: string | any[]): AssistantMessage {

 describe('anthropicMessagesToOpenAI', () => {
  test('converts system prompt to system message', () => {
-    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('hello')],
-      ['You are helpful.'] as any,
-    )
+    const result = anthropicMessagesToOpenAI([makeUserMsg('hello')], [
+      'You are helpful.',
+    ] as any)
    expect(result[0]).toEqual({ role: 'system', content: 'You are helpful.' })
  })

  test('joins multiple system prompt strings', () => {
-    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('hi')],
-      ['Part 1', 'Part 2'] as any,
-    )
+    const result = anthropicMessagesToOpenAI([makeUserMsg('hi')], [
+      'Part 1',
+      'Part 2',
+    ] as any)
    expect(result[0]).toEqual({ role: 'system', content: 'Part 1\n\nPart 2' })
  })

  test('skips empty system prompt', () => {
-    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('hi')],
-      [] as any,
-    )
+    const result = anthropicMessagesToOpenAI([makeUserMsg('hi')], [] as any)
    expect(result[0].role).toBe('user')
  })

@@ -54,10 +50,12 @@ describe('anthropicMessagesToOpenAI', () => {

  test('converts user message with content array', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        { type: 'text', text: 'line 1' },
-        { type: 'text', text: 'line 2' },
-      ])],
+      [
+        makeUserMsg([
+          { type: 'text', text: 'line 1' },
+          { type: 'text', text: 'line 2' },
+        ]),
+      ],
      [] as any,
    )
    expect(result).toEqual([{ role: 'user', content: 'line 1\nline 2' }])
@@ -73,52 +71,64 @@ describe('anthropicMessagesToOpenAI', () => {

  test('converts assistant message with tool_use', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeAssistantMsg([
-        { type: 'text', text: 'Let me help.' },
-        {
-          type: 'tool_use' as const,
-          id: 'toolu_123',
-          name: 'bash',
-          input: { command: 'ls' },
-        },
-      ])],
+      [
+        makeAssistantMsg([
+          { type: 'text', text: 'Let me help.' },
+          {
+            type: 'tool_use' as const,
+            id: 'toolu_123',
+            name: 'bash',
+            input: { command: 'ls' },
+          },
+        ]),
+      ],
      [] as any,
    )
-    expect(result).toEqual([{
-      role: 'assistant',
-      content: 'Let me help.',
-      tool_calls: [{
-        id: 'toolu_123',
-        type: 'function',
-        function: { name: 'bash', arguments: '{"command":"ls"}' },
-      }],
-    }])
+    expect(result).toEqual([
+      {
+        role: 'assistant',
+        content: 'Let me help.',
+        tool_calls: [
+          {
+            id: 'toolu_123',
+            type: 'function',
+            function: { name: 'bash', arguments: '{"command":"ls"}' },
+          },
+        ],
+      },
+    ])
  })

  test('converts tool_result to tool message', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        {
-          type: 'tool_result' as const,
-          tool_use_id: 'toolu_123',
-          content: 'file1.txt\nfile2.txt',
-        },
-      ])],
+      [
+        makeUserMsg([
+          {
+            type: 'tool_result' as const,
+            tool_use_id: 'toolu_123',
+            content: 'file1.txt\nfile2.txt',
+          },
+        ]),
+      ],
      [] as any,
    )
-    expect(result).toEqual([{
-      role: 'tool',
-      tool_call_id: 'toolu_123',
-      content: 'file1.txt\nfile2.txt',
-    }])
+    expect(result).toEqual([
+      {
+        role: 'tool',
+        tool_call_id: 'toolu_123',
+        content: 'file1.txt\nfile2.txt',
+      },
+    ])
  })

  test('strips thinking blocks', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeAssistantMsg([
-        { type: 'thinking' as const, thinking: 'internal thoughts...' },
-        { type: 'text', text: 'visible response' },
-      ])],
+      [
+        makeAssistantMsg([
+          { type: 'thinking' as const, thinking: 'internal thoughts...' },
+          { type: 'text', text: 'visible response' },
+        ]),
+      ],
      [] as any,
    )
    expect(result).toEqual([{ role: 'assistant', content: 'visible response' }])
@@ -157,91 +167,105 @@ describe('anthropicMessagesToOpenAI', () => {

  test('converts base64 image to image_url', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        { type: 'text', text: 'what is this?' },
-        {
-          type: 'image' as const,
-          source: {
-            type: 'base64',
-            media_type: 'image/png',
-            data: 'iVBORw0KGgo=',
+      [
+        makeUserMsg([
+          { type: 'text', text: 'what is this?' },
+          {
+            type: 'image' as const,
+            source: {
+              type: 'base64',
+              media_type: 'image/png',
+              data: 'iVBORw0KGgo=',
+            },
          },
-        },
-      ])],
+        ]),
+      ],
      [] as any,
    )
-    expect(result).toEqual([{
-      role: 'user',
-      content: [
-        { type: 'text', text: 'what is this?' },
-        {
-          type: 'image_url',
-          image_url: { url: 'data:image/png;base64,iVBORw0KGgo=' },
-        },
-      ],
-    }])
+    expect(result).toEqual([
+      {
+        role: 'user',
+        content: [
+          { type: 'text', text: 'what is this?' },
+          {
+            type: 'image_url',
+            image_url: { url: 'data:image/png;base64,iVBORw0KGgo=' },
+          },
+        ],
+      },
+    ])
  })

  test('converts url image to image_url', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        {
-          type: 'image' as const,
-          source: {
-            type: 'url',
-            url: 'https://example.com/img.png',
+      [
+        makeUserMsg([
+          {
+            type: 'image' as const,
+            source: {
+              type: 'url',
+              url: 'https://example.com/img.png',
+            },
          },
-        },
-      ])],
+        ]),
+      ],
      [] as any,
    )
-    expect(result).toEqual([{
-      role: 'user',
-      content: [
-        {
-          type: 'image_url',
-          image_url: { url: 'https://example.com/img.png' },
-        },
-      ],
-    }])
+    expect(result).toEqual([
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image_url',
+            image_url: { url: 'https://example.com/img.png' },
+          },
+        ],
+      },
+    ])
  })

  test('converts image-only message without text', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        {
-          type: 'image' as const,
-          source: {
-            type: 'base64',
-            media_type: 'image/jpeg',
-            data: '/9j/4AAQ',
+      [
+        makeUserMsg([
+          {
+            type: 'image' as const,
+            source: {
+              type: 'base64',
+              media_type: 'image/jpeg',
+              data: '/9j/4AAQ',
+            },
          },
-        },
-      ])],
+        ]),
+      ],
      [] as any,
    )
-    expect(result).toEqual([{
-      role: 'user',
-      content: [
-        {
-          type: 'image_url',
-          image_url: { url: 'data:image/jpeg;base64,/9j/4AAQ' },
-        },
-      ],
-    }])
+    expect(result).toEqual([
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'image_url',
+            image_url: { url: 'data:image/jpeg;base64,/9j/4AAQ' },
+          },
+        ],
+      },
+    ])
  })

  test('defaults to image/png when media_type is missing', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg([
-        {
-          type: 'image' as const,
-          source: {
-            type: 'base64',
-            data: 'ABC123',
+      [
+        makeUserMsg([
+          {
+            type: 'image' as const,
+            source: {
+              type: 'base64',
+              data: 'ABC123',
+            },
          },
-        },
-      ])],
+        ]),
+      ],
      [] as any,
    )
    expect((result[0].content as any[])[0].image_url.url).toBe(
@@ -253,10 +277,16 @@ describe('anthropicMessagesToOpenAI', () => {
 describe('DeepSeek thinking mode (enableThinking)', () => {
  test('preserves thinking block as reasoning_content when enabled', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('question'), makeAssistantMsg([
-        { type: 'thinking' as const, thinking: 'Let me reason about this...' },
-        { type: 'text', text: 'The answer is 42.' },
-      ])],
+      [
+        makeUserMsg('question'),
+        makeAssistantMsg([
+          {
+            type: 'thinking' as const,
+            thinking: 'Let me reason about this...',
+          },
+          { type: 'text', text: 'The answer is 42.' },
+        ]),
+      ],
      [] as any,
      { enableThinking: true },
    )
@@ -271,10 +301,12 @@ describe('DeepSeek thinking mode (enableThinking)', () => {

  test('drops thinking block when enableThinking is false (default)', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeAssistantMsg([
-        { type: 'thinking' as const, thinking: 'internal thoughts...' },
-        { type: 'text', text: 'visible response' },
-      ])],
+      [
+        makeAssistantMsg([
+          { type: 'thinking' as const, thinking: 'internal thoughts...' },
+          { type: 'text', text: 'visible response' },
+        ]),
+      ],
      [] as any,
    )
    const assistant = result[0] as any
@@ -287,7 +319,10 @@ describe('DeepSeek thinking mode (enableThinking)', () => {
      [
        makeUserMsg('what is the weather?'),
        makeAssistantMsg([
-          { type: 'thinking' as const, thinking: 'I need to call the weather tool.' },
+          {
+            type: 'thinking' as const,
+            thinking: 'I need to call the weather tool.',
+          },
          { type: 'text', text: '' },
          {
            type: 'tool_use' as const,
@@ -399,18 +434,27 @@ describe('DeepSeek thinking mode (enableThinking)', () => {
    const assistants = result.filter(m => m.role === 'assistant')
    expect(assistants.length).toBe(3)
    // All iterations within the same turn preserve reasoning
-    expect((assistants[0] as any).reasoning_content).toBe('I need the date first.')
-    expect((assistants[1] as any).reasoning_content).toBe('Now I can get the weather.')
-    expect((assistants[2] as any).reasoning_content).toBe('I have the info now.')
+    expect((assistants[0] as any).reasoning_content).toBe(
+      'I need the date first.',
+    )
+    expect((assistants[1] as any).reasoning_content).toBe(
+      'Now I can get the weather.',
+    )
+    expect((assistants[2] as any).reasoning_content).toBe(
+      'I have the info now.',
+    )
  })

  test('handles multiple thinking blocks in single assistant message', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('question'), makeAssistantMsg([
-        { type: 'thinking' as const, thinking: 'First thought.' },
-        { type: 'thinking' as const, thinking: 'Second thought.' },
-        { type: 'text', text: 'Final answer.' },
-      ])],
+      [
+        makeUserMsg('question'),
+        makeAssistantMsg([
+          { type: 'thinking' as const, thinking: 'First thought.' },
+          { type: 'thinking' as const, thinking: 'Second thought.' },
+          { type: 'text', text: 'Final answer.' },
+        ]),
+      ],
      [] as any,
      { enableThinking: true },
    )
@@ -420,10 +464,13 @@ describe('DeepSeek thinking mode (enableThinking)', () => {

  test('skips empty thinking blocks', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('question'), makeAssistantMsg([
-        { type: 'thinking' as const, thinking: '' },
-        { type: 'text', text: 'Answer.' },
-      ])],
+      [
+        makeUserMsg('question'),
+        makeAssistantMsg([
+          { type: 'thinking' as const, thinking: '' },
+          { type: 'text', text: 'Answer.' },
+        ]),
+      ],
      [] as any,
      { enableThinking: true },
    )
@@ -481,15 +528,18 @@ describe('DeepSeek thinking mode (enableThinking)', () => {

  test('sets content to null when only thinking and tool_calls present', () => {
    const result = anthropicMessagesToOpenAI(
-      [makeUserMsg('question'), makeAssistantMsg([
-        { type: 'thinking' as const, thinking: 'Reasoning only.' },
-        {
-          type: 'tool_use' as const,
-          id: 'toolu_001',
-          name: 'bash',
-          input: { command: 'ls' },
-        },
-      ])],
+      [
+        makeUserMsg('question'),
+        makeAssistantMsg([
+          { type: 'thinking' as const, thinking: 'Reasoning only.' },
+          {
+            type: 'tool_use' as const,
+            id: 'toolu_001',
+            name: 'bash',
+            input: { command: 'ls' },
+          },
+        ]),
+      ],
      [] as any,
      { enableThinking: true },
    )
--- a/packages/@ant/model-provider/src/shared/tests/openaiConvertTools.test.ts
+++ b/packages/@ant/model-provider/src/shared/tests/openaiConvertTools.test.ts
@@ -18,25 +18,29 @@ describe('anthropicToolsToOpenAI', () => {

    const result = anthropicToolsToOpenAI(tools as any)

-    expect(result).toEqual([{
-      type: 'function',
-      function: {
-        name: 'bash',
-        description: 'Run a bash command',
-        parameters: {
-          type: 'object',
-          properties: { command: { type: 'string' } },
-          required: ['command'],
+    expect(result).toEqual([
+      {
+        type: 'function',
+        function: {
+          name: 'bash',
+          description: 'Run a bash command',
+          parameters: {
+            type: 'object',
+            properties: { command: { type: 'string' } },
+            required: ['command'],
+          },
        },
      },
-    }])
+    ])
  })

  test('uses empty schema when input_schema missing', () => {
    const tools = [{ type: 'custom', name: 'noop', description: 'no-op' }]
    const result = anthropicToolsToOpenAI(tools as any)

-    expect((result[0] as { function: { parameters: unknown } }).function.parameters).toEqual({ type: 'object', properties: {} })
+    expect(
+      (result[0] as { function: { parameters: unknown } }).function.parameters,
+    ).toEqual({ type: 'object', properties: {} })
  })

  test('strips Anthropic-specific fields', () => {
@@ -76,7 +80,8 @@ describe('anthropicToolsToOpenAI', () => {
      },
    ]
    const result = anthropicToolsToOpenAI(tools as any)
-    const props = (result[0] as { function: { parameters: any } }).function.parameters as any
+    const props = (result[0] as { function: { parameters: any } }).function
+      .parameters as any
    expect(props.properties.mode).toEqual({ enum: ['read'] })
    expect(props.properties.mode.const).toBeUndefined()
    expect(props.properties.name).toEqual({ type: 'string' })
@@ -110,8 +115,11 @@ describe('anthropicToolsToOpenAI', () => {
      },
    ]
    const result = anthropicToolsToOpenAI(tools as any)
-    const params = (result[0] as { function: { parameters: any } }).function.parameters as any
-    expect(params.properties.outer.properties.inner).toEqual({ enum: ['fixed'] })
+    const params = (result[0] as { function: { parameters: any } }).function
+      .parameters as any
+    expect(params.properties.outer.properties.inner).toEqual({
+      enum: ['fixed'],
+    })
    expect(params.definitions.MyType.properties.field).toEqual({ enum: [42] })
  })

@@ -125,18 +133,17 @@ describe('anthropicToolsToOpenAI', () => {
          type: 'object',
          properties: {
            val: {
-              anyOf: [
-                { const: 'a' },
-                { const: 'b' },
-                { type: 'string' },
-              ],
+              anyOf: [{ const: 'a' }, { const: 'b' }, { type: 'string' }],
            },
          },
        },
      },
    ]
    const result = anthropicToolsToOpenAI(tools as any)
-    const anyOf = ((result[0] as { function: { parameters: any } }).function.parameters as any).properties.val.anyOf
+    const anyOf = (
+      (result[0] as { function: { parameters: any } }).function
+        .parameters as any
+    ).properties.val.anyOf
    expect(anyOf[0]).toEqual({ enum: ['a'] })
    expect(anyOf[1]).toEqual({ enum: ['b'] })
    expect(anyOf[2]).toEqual({ type: 'string' })
--- a/packages/@ant/model-provider/src/shared/openaiConvertMessages.ts
+++ b/packages/@ant/model-provider/src/shared/openaiConvertMessages.ts
@@ -62,16 +62,18 @@ export function anthropicMessagesToOpenAI(
        // A user message starts a new turn if it contains any non-tool_result content
        // (text, image, or other media). Tool results alone do NOT start a new turn
        // because they are continuations of the previous assistant tool call.
-        const startsNewUserTurn = typeof content === 'string'
-          ? content.length > 0
-          : Array.isArray(content) && content.some(
-              (b: any) =>
-                typeof b === 'string' ||
-                (b &&
-                  typeof b === 'object' &&
-                  'type' in b &&
-                  b.type !== 'tool_result'),
-            )
+        const startsNewUserTurn =
+          typeof content === 'string'
+            ? content.length > 0
+            : Array.isArray(content) &&
+              content.some(
+                (b: any) =>
+                  typeof b === 'string' ||
+                  (b &&
+                    typeof b === 'object' &&
+                    'type' in b &&
+                    b.type !== 'tool_result'),
+              )
        if (startsNewUserTurn) {
          turnBoundaries.add(i)
        }
@@ -88,7 +90,8 @@ export function anthropicMessagesToOpenAI(
      case 'assistant':
        // Preserve reasoning_content unless we're before a turn boundary
        // (i.e., from a previous user Q&A round)
-        const preserveReasoning = enableThinking && !isBeforeAnyTurnBoundary(i, turnBoundaries)
+        const preserveReasoning =
+          enableThinking && !isBeforeAnyTurnBoundary(i, turnBoundaries)
        result.push(...convertInternalAssistantMessage(msg, preserveReasoning))
        break
      default:
@@ -101,9 +104,7 @@ export function anthropicMessagesToOpenAI(

 function systemPromptToText(systemPrompt: SystemPrompt): string {
  if (!systemPrompt || systemPrompt.length === 0) return ''
-  return systemPrompt
-    .filter(Boolean)
-    .join('\n\n')
+  return systemPrompt.filter(Boolean).join('\n\n')
 }

 /**
@@ -131,7 +132,8 @@ function convertInternalUserMessage(
  } else if (Array.isArray(content)) {
    const textParts: string[] = []
    const toolResults: BetaToolResultBlockParam[] = []
-    const imageParts: Array<{ type: 'image_url'; image_url: { url: string } }> = []
+    const imageParts: Array<{ type: 'image_url'; image_url: { url: string } }> =
+      []

    for (const block of content) {
      if (typeof block === 'string') {
@@ -141,7 +143,9 @@ function convertInternalUserMessage(
      } else if (block.type === 'tool_result') {
        toolResults.push(block as BetaToolResultBlockParam)
      } else if (block.type === 'image') {
-        const imagePart = convertImageBlockToOpenAI(block as unknown as Record<string, unknown>)
+        const imagePart = convertImageBlockToOpenAI(
+          block as unknown as Record<string, unknown>,
+        )
        if (imagePart) {
          imageParts.push(imagePart)
        }
@@ -158,7 +162,10 @@ function convertInternalUserMessage(

    // 如果有图片，构建多模态 content 数组
    if (imageParts.length > 0) {
-      const multiContent: Array<{ type: 'text'; text: string } | { type: 'image_url'; image_url: { url: string } }> = []
+      const multiContent: Array<
+        | { type: 'text'; text: string }
+        | { type: 'image_url'; image_url: { url: string } }
+      > = []
      if (textParts.length > 0) {
        multiContent.push({ type: 'text', text: textParts.join('\n') })
      }
@@ -229,7 +236,9 @@ function convertInternalAssistantMessage(
  }

  const textParts: string[] = []
-  const toolCalls: NonNullable<ChatCompletionAssistantMessageParam['tool_calls']> = []
+  const toolCalls: NonNullable<
+    ChatCompletionAssistantMessageParam['tool_calls']
+  > = []
  const reasoningParts: string[] = []

  for (const block of content) {
@@ -250,7 +259,8 @@ function convertInternalAssistantMessage(
      })
    } else if (block.type === 'thinking' && preserveReasoning) {
      // DeepSeek thinking mode: preserve reasoning_content for tool call iterations
-      const thinkingText = (block as unknown as Record<string, unknown>).thinking
+      const thinkingText = (block as unknown as Record<string, unknown>)
+        .thinking
      if (typeof thinkingText === 'string' && thinkingText) {
        reasoningParts.push(thinkingText)
      }
@@ -262,7 +272,9 @@ function convertInternalAssistantMessage(
    role: 'assistant',
    content: textParts.length > 0 ? textParts.join('\n') : null,
    ...(toolCalls.length > 0 && { tool_calls: toolCalls }),
-    ...(reasoningParts.length > 0 && { reasoning_content: reasoningParts.join('\n') }),
+    ...(reasoningParts.length > 0 && {
+      reasoning_content: reasoningParts.join('\n'),
+    }),
  }

  return [result]
--- a/packages/@ant/model-provider/src/shared/openaiConvertTools.ts
+++ b/packages/@ant/model-provider/src/shared/openaiConvertTools.ts
@@ -16,21 +16,27 @@ export function anthropicToolsToOpenAI(
    .filter(tool => {
      // Only convert standard tools (skip server tools like computer_use, etc.)
      const toolType = (tool as unknown as { type?: string }).type
-      return tool.type === 'custom' || !('type' in tool) || toolType !== 'server'
+      return (
+        tool.type === 'custom' || !('type' in tool) || toolType !== 'server'
+      )
    })
    .map(tool => {
      // Handle the various tool shapes from Anthropic SDK
      const anyTool = tool as unknown as Record<string, unknown>
      const name = (anyTool.name as string) || ''
      const description = (anyTool.description as string) || ''
-      const inputSchema = anyTool.input_schema as Record<string, unknown> | undefined
+      const inputSchema = anyTool.input_schema as
+        | Record<string, unknown>
+        | undefined

      return {
        type: 'function' as const,
        function: {
          name,
          description,
-          parameters: sanitizeJsonSchema(inputSchema || { type: 'object', properties: {} }),
+          parameters: sanitizeJsonSchema(
+            inputSchema || { type: 'object', properties: {} },
+          ),
        },
      } satisfies ChatCompletionTool
    })
@@ -43,7 +49,9 @@ export function anthropicToolsToOpenAI(
 * support the `const` keyword in JSON Schema. Convert it to `enum` with a
 * single-element array, which is semantically equivalent.
 */
-function sanitizeJsonSchema(schema: Record<string, unknown>): Record<string, unknown> {
+function sanitizeJsonSchema(
+  schema: Record<string, unknown>,
+): Record<string, unknown> {
  if (!schema || typeof schema !== 'object') return schema

  const result = { ...schema }
@@ -55,20 +63,37 @@ function sanitizeJsonSchema(schema: Record<string, unknown>): Record<string, unk
  }

  // Recursively process nested schemas
-  const objectKeys = ['properties', 'definitions', '$defs', 'patternProperties'] as const
+  const objectKeys = [
+    'properties',
+    'definitions',
+    '$defs',
+    'patternProperties',
+  ] as const
  for (const key of objectKeys) {
    const nested = result[key]
    if (nested && typeof nested === 'object') {
      const sanitized: Record<string, unknown> = {}
      for (const [k, v] of Object.entries(nested as Record<string, unknown>)) {
-        sanitized[k] = v && typeof v === 'object' ? sanitizeJsonSchema(v as Record<string, unknown>) : v
+        sanitized[k] =
+          v && typeof v === 'object'
+            ? sanitizeJsonSchema(v as Record<string, unknown>)
+            : v
      }
      result[key] = sanitized
    }
  }

  // Recursively process single-schema keys
-  const singleKeys = ['items', 'additionalProperties', 'not', 'if', 'then', 'else', 'contains', 'propertyNames'] as const
+  const singleKeys = [
+    'items',
+    'additionalProperties',
+    'not',
+    'if',
+    'then',
+    'else',
+    'contains',
+    'propertyNames',
+  ] as const
  for (const key of singleKeys) {
    const nested = result[key]
    if (nested && typeof nested === 'object' && !Array.isArray(nested)) {
@@ -82,7 +107,9 @@ function sanitizeJsonSchema(schema: Record<string, unknown>): Record<string, unk
    const nested = result[key]
    if (Array.isArray(nested)) {
      result[key] = nested.map(item =>
-        item && typeof item === 'object' ? sanitizeJsonSchema(item as Record<string, unknown>) : item
+        item && typeof item === 'object'
+          ? sanitizeJsonSchema(item as Record<string, unknown>)
+          : item,
      )
    }
  }
--- a/packages/@ant/model-provider/src/shared/openaiStreamAdapter.ts
+++ b/packages/@ant/model-provider/src/shared/openaiStreamAdapter.ts
@@ -42,7 +42,10 @@ export async function* adaptOpenAIStreamToAnthropic(
  let currentContentIndex = -1

  // Track tool_use blocks: tool_calls index → { contentIndex, id, name, arguments }
-  const toolBlocks = new Map<number, { contentIndex: number; id: string; name: string; arguments: string }>()
+  const toolBlocks = new Map<
+    number,
+    { contentIndex: number; id: string; name: string; arguments: string }
+  >()

  // Track thinking block state
  let thinkingBlockOpen = false
@@ -197,7 +200,8 @@ export async function* adaptOpenAIStreamToAnthropic(

          // Start new tool_use block
          currentContentIndex++
-          const toolId = tc.id || `toolu_${randomUUID().replace(/-/g, '').slice(0, 24)}`
+          const toolId =
+            tc.id || `toolu_${randomUUID().replace(/-/g, '').slice(0, 24)}`
          const toolName = tc.function?.name || ''

          toolBlocks.set(tcIndex, {
--- a/packages/builtin-tools/src/tools/AgentTool/tests/agentToolUtils.test.ts
+++ b/packages/builtin-tools/src/tools/AgentTool/tests/agentToolUtils.test.ts
@@ -1,4 +1,5 @@
 import { mock, describe, expect, test } from "bun:test";
+import { debugMock } from "../../../../../../tests/mocks/debug";

 // ─── Mocks for agentToolUtils.ts dependencies ───
 // Only mock modules that are truly unavailable or cause side effects.
@@ -87,20 +88,7 @@ mock.module("src/tasks/LocalAgentTask/LocalAgentTask.js", () => ({
  updateProgressFromMessage: noop,
 }));

-mock.module("src/utils/debug.ts", () => ({
-  getMinDebugLogLevel: () => "warn",
-  isDebugMode: () => false,
-  enableDebugLogging: () => false,
-  getDebugFilter: () => null,
-  isDebugToStdErr: () => false,
-  getDebugFilePath: () => null,
-  setHasFormattedOutput: noop,
-  getHasFormattedOutput: () => false,
-  flushDebugLogs: async () => {},
-  logForDebugging: noop,
-  getDebugLogPath: () => "",
-  logAntError: noop,
-}));
+mock.module("src/utils/debug.ts", debugMock);

 mock.module("src/utils/errors.js", () => ({
  ClaudeError: class extends Error {},
--- a/packages/builtin-tools/src/tools/CtxInspectTool/CtxInspectTool.ts
+++ b/packages/builtin-tools/src/tools/CtxInspectTool/CtxInspectTool.ts
@@ -2,6 +2,12 @@ import { z } from 'zod/v4'
 import type { ToolResultBlockParam } from 'src/Tool.js'
 import { buildTool } from 'src/Tool.js'
 import { lazySchema } from 'src/utils/lazySchema.js'
+import { tokenCountWithEstimation } from 'src/utils/tokens.js'
+import {
+  getStats,
+  isContextCollapseEnabled,
+} from 'src/services/contextCollapse/index.js'
+import { isSessionMemoryInitialized } from 'src/services/SessionMemory/sessionMemoryUtils.js'

 const CTX_INSPECT_TOOL_NAME = 'CtxInspect'

@@ -19,6 +25,10 @@ type CtxInput = z.infer<InputSchema>
 type CtxOutput = {
  total_tokens: number
  message_count: number
+  context_window_model: string
+  prompt_caching_enabled: boolean
+  session_memory_enabled: boolean
+  context_collapse_enabled: boolean
  summary: string
 }

@@ -67,13 +77,45 @@ Use this to understand your context budget before deciding whether to snip old m
    }
  },

-  async call() {
-    // Context inspection is wired into the context collapse system.
+  async call(input: CtxInput, context) {
+    const messages = context.messages ?? []
+    const model = context.options?.mainLoopModel ?? 'unknown'
+    const totalTokens = tokenCountWithEstimation(messages)
+    const collapseEnabled = isContextCollapseEnabled()
+    const collapseStats = getStats()
+    const focused = input.query?.trim()
+
+    const sessionMemoryEnabled = isSessionMemoryInitialized()
+    // Prompt caching is an API-level feature controlled by the provider, not
+    // a user-facing toggle. Report as enabled only for providers known to
+    // support Anthropic-style prompt caching (first-party, Bedrock, Vertex).
+    const promptCachingEnabled = !model.startsWith('openai/') &&
+      !model.startsWith('grok/') &&
+      !model.startsWith('gemini/')
+
+    const summaryParts = [
+      focused ? `Focus: ${focused}` : 'Overall context summary',
+      `Model context: ${model}`,
+      `Prompt caching: ${promptCachingEnabled ? 'enabled' : 'disabled'}`,
+      `Session memory: ${sessionMemoryEnabled ? 'enabled' : 'disabled'}`,
+      `Context collapse: ${collapseEnabled ? 'enabled' : 'disabled'}`,
+    ]
+
+    if (collapseEnabled) {
+      summaryParts.push(
+        `Collapse spans: ${collapseStats.collapsedSpans} committed, ${collapseStats.stagedSpans} staged, ${collapseStats.collapsedMessages} messages summarized`,
+      )
+    }
+
    return {
      data: {
-        total_tokens: 0,
-        message_count: 0,
-        summary: 'Context inspection requires the CONTEXT_COLLAPSE runtime.',
+        total_tokens: totalTokens,
+        message_count: messages.length,
+        context_window_model: model,
+        prompt_caching_enabled: promptCachingEnabled,
+        session_memory_enabled: sessionMemoryEnabled,
+        context_collapse_enabled: collapseEnabled,
+        summary: summaryParts.join('\n'),
      },
    }
  },
--- a/packages/builtin-tools/src/tools/CtxInspectTool/tests/CtxInspectTool.test.ts
+++ b/packages/builtin-tools/src/tools/CtxInspectTool/tests/CtxInspectTool.test.ts
@@ -0,0 +1,202 @@
+import { afterEach, beforeEach, describe, expect, mock, test } from 'bun:test'
+import { logMock } from '../../../../../../tests/mocks/log'
+
+mock.module('src/utils/log.ts', logMock)
+
+mock.module('src/services/tokenEstimation.ts', () => ({
+  roughTokenCountEstimation: (text: string) => Math.ceil(text.length / 4),
+  roughTokenCountEstimationForMessages: (msgs: unknown[]) => msgs.length * 64,
+  roughTokenCountEstimationForMessage: () => 64,
+  roughTokenCountEstimationForFileType: () => 64,
+  bytesPerTokenForFileType: () => 4,
+  countTokensWithAPI: async () => 0,
+  countMessagesTokensWithAPI: async () => 0,
+  countTokensViaHaikuFallback: async () => 0,
+}))
+
+let sessionMemoryInitialized = false
+mock.module('src/services/SessionMemory/sessionMemoryUtils.ts', () => ({
+  isSessionMemoryInitialized: () => sessionMemoryInitialized,
+  waitForSessionMemoryExtraction: async () => {},
+  getLastSummarizedMessageId: () => undefined,
+  getSessionMemoryContent: async () => null,
+  setLastSummarizedMessageId: () => {},
+  markExtractionStarted: () => {},
+  markExtractionCompleted: () => {},
+  setSessionMemoryConfig: () => {},
+  getSessionMemoryConfig: () => ({}),
+  recordExtractionTokenCount: () => {},
+  markSessionMemoryInitialized: () => {},
+  hasMetInitializationThreshold: () => false,
+  hasMetUpdateThreshold: () => false,
+  getToolCallsBetweenUpdates: () => 0,
+  resetSessionMemoryState: () => {},
+  DEFAULT_SESSION_MEMORY_CONFIG: {},
+}))
+
+mock.module('src/utils/slowOperations.ts', () => ({
+  jsonStringify: JSON.stringify,
+  jsonParse: JSON.parse,
+  slowLogging: { enabled: false },
+  clone: (value: unknown) => structuredClone(value),
+  cloneDeep: (value: unknown) => structuredClone(value),
+  callerFrame: () => '',
+  SLOW_OPERATION_THRESHOLD_MS: 100,
+  writeFileSync_DEPRECATED: () => {},
+}))
+
+const { initContextCollapse, resetContextCollapse } = await import(
+  'src/services/contextCollapse/index.js'
+)
+const { tokenCountWithEstimation } = await import('src/utils/tokens.js')
+const { CtxInspectTool } = await import('../CtxInspectTool.js')
+
+function makeUserMessage(text: string) {
+  return {
+    type: 'user' as const,
+    uuid: `user-${text}`,
+    message: { role: 'user' as const, content: text },
+  }
+}
+
+function makeAssistantMessage(text: string) {
+  return {
+    type: 'assistant' as const,
+    uuid: `assistant-${text}`,
+    message: {
+      role: 'assistant' as const,
+      content: [{ type: 'text' as const, text }],
+    },
+  }
+}
+
+function makeContext(messages: unknown[], mainLoopModel = 'claude-sonnet-4-6') {
+  return {
+    messages,
+    options: {
+      mainLoopModel,
+    },
+    getAppState: () => ({}),
+  } as any
+}
+
+const allowTool = async (input: Record<string, unknown>) => ({
+  behavior: 'allow' as const,
+  updatedInput: input,
+})
+
+const parentMessage = makeAssistantMessage('Parent tool call')
+
+beforeEach(() => {
+  resetContextCollapse()
+  sessionMemoryInitialized = false
+})
+
+afterEach(() => {
+  resetContextCollapse()
+  sessionMemoryInitialized = false
+})
+
+describe('CtxInspectTool', () => {
+  test('tool exports and metadata remain stable', async () => {
+    expect(CtxInspectTool).toBeDefined()
+    expect(CtxInspectTool.name).toBe('CtxInspect')
+    expect(typeof CtxInspectTool.call).toBe('function')
+    expect(await CtxInspectTool.description()).toContain('context')
+    expect(CtxInspectTool.userFacingName()).toBe('CtxInspect')
+    expect(CtxInspectTool.isReadOnly()).toBe(true)
+    expect(CtxInspectTool.isConcurrencySafe()).toBe(true)
+  })
+
+  test('formats tool results for transcript rendering', () => {
+    const block = CtxInspectTool.mapToolResultToToolResultBlockParam(
+      {
+        total_tokens: 192,
+        message_count: 3,
+        context_window_model: 'claude-sonnet-4-6',
+        prompt_caching_enabled: true,
+        session_memory_enabled: true,
+        context_collapse_enabled: false,
+        summary: 'Context collapse: disabled',
+      },
+      'tool-use-id',
+    )
+
+    expect(block.tool_use_id).toBe('tool-use-id')
+    expect(block.content).toContain('192 tokens')
+    expect(block.content).toContain('3 messages')
+    expect(block.content).toContain('Context collapse: disabled')
+  })
+
+  test('returns live context counts and mechanism state', async () => {
+    const messages = [
+      makeUserMessage('Inspect the current context budget.'),
+      makeAssistantMessage('Looking at the current conversation state.'),
+    ]
+    const context = makeContext(messages, 'claude-sonnet-4-6')
+
+    const result = await (CtxInspectTool as any).call(
+      {},
+      context,
+      allowTool,
+      parentMessage,
+    )
+
+    expect(Object.keys(result.data).sort()).toEqual([
+      'context_collapse_enabled',
+      'context_window_model',
+      'message_count',
+      'prompt_caching_enabled',
+      'session_memory_enabled',
+      'summary',
+      'total_tokens',
+    ])
+    expect(result.data.message_count).toBe(messages.length)
+    expect(result.data.total_tokens).toBe(tokenCountWithEstimation(messages as any))
+    expect(result.data.context_window_model).toBe('claude-sonnet-4-6')
+    expect(result.data.prompt_caching_enabled).toBe(true)
+    expect(result.data.session_memory_enabled).toBe(false)
+    expect(result.data.context_collapse_enabled).toBe(false)
+    expect(result.data.summary).toContain('Overall context summary')
+    expect(result.data.summary).toContain('Session memory: disabled')
+    expect(result.data.summary).toContain('Context collapse: disabled')
+  })
+
+  test('query input focuses summary and collapse runtime changes the reported state', async () => {
+    const messages = [
+      makeUserMessage('Show me tool usage pressure in this thread.'),
+      makeAssistantMessage('Summarizing tool-heavy context now.'),
+    ]
+    const context = makeContext(messages, 'claude-sonnet-4-6')
+
+    const disabledResult = await (CtxInspectTool as any).call(
+      { query: 'tool usage' },
+      context,
+      allowTool,
+      parentMessage,
+    )
+
+    initContextCollapse()
+
+    const enabledResult = await (CtxInspectTool as any).call(
+      { query: 'tool usage' },
+      context,
+      allowTool,
+      parentMessage,
+    )
+
+    expect(disabledResult.data.message_count).toBe(messages.length)
+    expect(enabledResult.data.message_count).toBe(messages.length)
+    expect(disabledResult.data.total_tokens).toBe(
+      tokenCountWithEstimation(messages as any),
+    )
+    expect(enabledResult.data.total_tokens).toBe(
+      tokenCountWithEstimation(messages as any),
+    )
+    expect(disabledResult.data.summary).toContain('Focus: tool usage')
+    expect(disabledResult.data.context_collapse_enabled).toBe(false)
+    expect(enabledResult.data.context_collapse_enabled).toBe(true)
+    expect(enabledResult.data.summary).toContain('Context collapse: enabled')
+    expect(enabledResult.data.summary).toContain('Collapse spans:')
+  })
+})
--- a/packages/builtin-tools/src/tools/DiscoverSkillsTool/DiscoverSkillsTool.ts
+++ b/packages/builtin-tools/src/tools/DiscoverSkillsTool/DiscoverSkillsTool.ts
@@ -0,0 +1,107 @@
+import { z } from 'zod/v4'
+import type { ToolResultBlockParam } from 'src/Tool.js'
+import { buildTool } from 'src/Tool.js'
+import { lazySchema } from 'src/utils/lazySchema.js'
+import {
+  DISCOVER_SKILLS_TOOL_NAME,
+  DESCRIPTION,
+  DISCOVER_SKILLS_PROMPT,
+} from './prompt.js'
+
+const inputSchema = lazySchema(() =>
+  z.strictObject({
+    description: z
+      .string()
+      .describe(
+        'Description of what you want to do. Be specific — e.g. "deploy a Next.js app to Cloudflare Workers" rather than just "deploy".',
+      ),
+    limit: z
+      .number()
+      .optional()
+      .describe('Maximum number of results to return (default: 5)'),
+  }),
+)
+type InputSchema = ReturnType<typeof inputSchema>
+type DiscoverInput = z.infer<InputSchema>
+
+type DiscoverOutput = {
+  results: Array<{ name: string; description: string; score: number }>
+  count: number
+}
+
+export const DiscoverSkillsTool = buildTool({
+  name: DISCOVER_SKILLS_TOOL_NAME,
+  searchHint: 'find search discover skills commands tools capabilities',
+  maxResultSizeChars: 10_000,
+  strict: true,
+
+  get inputSchema(): InputSchema {
+    return inputSchema()
+  },
+
+  async description() {
+    return DESCRIPTION
+  },
+  async prompt() {
+    return DISCOVER_SKILLS_PROMPT
+  },
+
+  isConcurrencySafe() {
+    return true
+  },
+  isReadOnly() {
+    return true
+  },
+
+  userFacingName() {
+    return 'Discover Skills'
+  },
+
+  renderToolUseMessage(input: Partial<DiscoverInput>) {
+    return `Searching skills: ${input.description?.slice(0, 80) ?? '...'}`
+  },
+
+  mapToolResultToToolResultBlockParam(
+    content: DiscoverOutput,
+    toolUseID: string,
+  ): ToolResultBlockParam {
+    if (content.count === 0) {
+      return {
+        tool_use_id: toolUseID,
+        type: 'tool_result',
+        content: 'No matching skills found for that description.',
+      }
+    }
+    const lines = content.results.map(
+      (r, i) =>
+        `${i + 1}. **${r.name}** (score: ${r.score.toFixed(2)})\n   ${r.description}`,
+    )
+    return {
+      tool_use_id: toolUseID,
+      type: 'tool_result',
+      content: `Found ${content.count} relevant skill(s):\n\n${lines.join('\n\n')}`,
+    }
+  },
+
+  async call(input: DiscoverInput, context) {
+    const { getSkillIndex, searchSkills } = await import(
+      'src/services/skillSearch/localSearch.js'
+    )
+    const { getCwd } = await import('src/utils/cwd.js')
+    const cwd = getCwd()
+
+    const index = await getSkillIndex(cwd)
+    const results = searchSkills(input.description, index, input.limit ?? 5)
+
+    return {
+      data: {
+        results: results.map(r => ({
+          name: r.name,
+          description: r.description,
+          score: r.score,
+        })),
+        count: results.length,
+      },
+    }
+  },
+})
--- a/packages/builtin-tools/src/tools/DiscoverSkillsTool/tests/DiscoverSkillsTool.test.ts
+++ b/packages/builtin-tools/src/tools/DiscoverSkillsTool/tests/DiscoverSkillsTool.test.ts
@@ -0,0 +1,54 @@
+import { describe, test, expect } from 'bun:test'
+import { DISCOVER_SKILLS_TOOL_NAME } from '../prompt.js'
+
+describe('DiscoverSkillsTool', () => {
+  test('DISCOVER_SKILLS_TOOL_NAME is not empty', () => {
+    expect(DISCOVER_SKILLS_TOOL_NAME).toBe('DiscoverSkills')
+    expect(DISCOVER_SKILLS_TOOL_NAME.length).toBeGreaterThan(0)
+  })
+
+  test('tool exports are functions', async () => {
+    const { DiscoverSkillsTool } = await import('../DiscoverSkillsTool.js')
+    expect(DiscoverSkillsTool).toBeDefined()
+    expect(DiscoverSkillsTool.name).toBe('DiscoverSkills')
+    expect(typeof DiscoverSkillsTool.call).toBe('function')
+  })
+
+  test('tool has correct metadata', async () => {
+    const { DiscoverSkillsTool } = await import('../DiscoverSkillsTool.js')
+    expect(await DiscoverSkillsTool.description()).toContain('skill')
+    expect(DiscoverSkillsTool.userFacingName()).toBe('Discover Skills')
+    expect(DiscoverSkillsTool.isReadOnly()).toBe(true)
+    expect(DiscoverSkillsTool.isConcurrencySafe()).toBe(true)
+  })
+
+  test('renderToolUseMessage formats input', async () => {
+    const { DiscoverSkillsTool } = await import('../DiscoverSkillsTool.js')
+    const msg = DiscoverSkillsTool.renderToolUseMessage({
+      description: 'deploy to cloudflare',
+    })
+    expect(msg).toContain('deploy to cloudflare')
+  })
+
+  test('mapToolResultToToolResultBlockParam formats empty results', async () => {
+    const { DiscoverSkillsTool } = await import('../DiscoverSkillsTool.js')
+    const result = DiscoverSkillsTool.mapToolResultToToolResultBlockParam(
+      { results: [], count: 0 },
+      'test-id',
+    )
+    expect(result.content).toContain('No matching skills')
+  })
+
+  test('mapToolResultToToolResultBlockParam formats results', async () => {
+    const { DiscoverSkillsTool } = await import('../DiscoverSkillsTool.js')
+    const result = DiscoverSkillsTool.mapToolResultToToolResultBlockParam(
+      {
+        results: [{ name: 'test-skill', description: 'A test skill', score: 0.85 }],
+        count: 1,
+      },
+      'test-id',
+    )
+    expect(result.content).toContain('test-skill')
+    expect(result.content).toContain('0.85')
+  })
+})
--- a/packages/builtin-tools/src/tools/DiscoverSkillsTool/prompt.ts
+++ b/packages/builtin-tools/src/tools/DiscoverSkillsTool/prompt.ts
@@ -1,3 +1,13 @@
-// Auto-generated stub — replace with real implementation
-export {};
-export const DISCOVER_SKILLS_TOOL_NAME: string = '';
+export const DISCOVER_SKILLS_TOOL_NAME = 'DiscoverSkills'
+
+export const DESCRIPTION =
+  'Search for relevant skills by describing what you want to do'
+
+export const DISCOVER_SKILLS_PROMPT = `Search for skills relevant to a task description. Returns matching skills ranked by relevance.
+
+Use this when:
+- The auto-surfaced skills don't cover your current task
+- You're pivoting to a different kind of work mid-conversation
+- You want to find specialized skills for an unusual workflow
+
+The search uses TF-IDF keyword matching against all registered skills (bundled, user-defined, and MCP-provided). Results include skill name, description, and relevance score.`
--- a/packages/builtin-tools/src/tools/FileEditTool/tests/utils.test.ts
+++ b/packages/builtin-tools/src/tools/FileEditTool/tests/utils.test.ts
@@ -1,22 +1,8 @@
 import { mock, describe, expect, test } from "bun:test";
+import { logMock } from "../../../../../../tests/mocks/log";

 // Mock log.ts to cut the heavy dependency chain
-mock.module("src/utils/log.ts", () => ({
-  logError: () => {},
-  logToFile: () => {},
-  getLogDisplayTitle: () => "",
-  logEvent: () => {},
-  logMCPError: () => {},
-  logMCPDebug: () => {},
-  dateToFilename: (d: Date) => d.toISOString().replace(/[:.]/g, "-"),
-  getLogFilePath: () => "/tmp/mock-log",
-  attachErrorLogSink: () => {},
-  getInMemoryErrors: () => [],
-  loadErrorLogs: async () => [],
-  getErrorLogByIndex: async () => null,
-  captureAPIRequest: () => {},
-  _resetErrorLogForTesting: () => {},
-}));
+mock.module("src/utils/log.ts", logMock);

 const {
  normalizeQuotes,
--- a/packages/builtin-tools/src/tools/LSPTool/tests/formatters.test.ts
+++ b/packages/builtin-tools/src/tools/LSPTool/tests/formatters.test.ts
@@ -1,9 +1,7 @@
 import { mock, describe, expect, test } from "bun:test";
+import { debugMock } from "../../../../../../tests/mocks/debug";

-mock.module("src/utils/debug.ts", () => ({
-  logForDebugging: () => {},
-  isDebugMode: () => false,
-}));
+mock.module("src/utils/debug.ts", debugMock);

 const {
  formatGoToDefinitionResult,
--- a/packages/builtin-tools/src/tools/RemoteTriggerTool/RemoteTriggerTool.ts
+++ b/packages/builtin-tools/src/tools/RemoteTriggerTool/RemoteTriggerTool.ts
@@ -11,6 +11,7 @@ import {
  getClaudeAIOAuthTokens,
 } from 'src/utils/auth.js'
 import { lazySchema } from 'src/utils/lazySchema.js'
+import { appendRemoteTriggerAuditRecord } from 'src/utils/remoteTriggerAudit.js'
 import { jsonStringify } from 'src/utils/slowOperations.js'
 import { DESCRIPTION, PROMPT, REMOTE_TRIGGER_TOOL_NAME } from './prompt.js'
 import { renderToolResultMessage, renderToolUseMessage } from './UI.js'
@@ -36,6 +37,7 @@ const outputSchema = lazySchema(() =>
  z.object({
    status: z.number(),
    json: z.string(),
+    audit_id: z.string().optional(),
  }),
 )
 type OutputSchema = ReturnType<typeof outputSchema>
@@ -76,77 +78,96 @@ export const RemoteTriggerTool = buildTool({
    return PROMPT
  },
  async call(input: Input, context: ToolUseContext) {
-    await checkAndRefreshOAuthTokenIfNeeded()
-    const accessToken = getClaudeAIOAuthTokens()?.accessToken
-    if (!accessToken) {
-      throw new Error(
-        'Not authenticated with a claude.ai account. Run /login and try again.',
-      )
-    }
-    const orgUUID = await getOrganizationUUID()
-    if (!orgUUID) {
-      throw new Error('Unable to resolve organization UUID.')
+    const auditBase = {
+      action: input.action,
+      ...(input.trigger_id ? { triggerId: input.trigger_id } : {}),
    }
+    try {
+      await checkAndRefreshOAuthTokenIfNeeded()
+      const accessToken = getClaudeAIOAuthTokens()?.accessToken
+      if (!accessToken) {
+        throw new Error(
+          'Not authenticated with a claude.ai account. Run /login and try again.',
+        )
+      }
+      const orgUUID = await getOrganizationUUID()
+      if (!orgUUID) {
+        throw new Error('Unable to resolve organization UUID.')
+      }

-    const base = `${getOauthConfig().BASE_API_URL}/v1/code/triggers`
-    const headers = {
-      Authorization: `Bearer ${accessToken}`,
-      'Content-Type': 'application/json',
-      'anthropic-version': '2023-06-01',
-      'anthropic-beta': TRIGGERS_BETA,
-      'x-organization-uuid': orgUUID,
-    }
+      const base = `${getOauthConfig().BASE_API_URL}/v1/code/triggers`
+      const headers = {
+        Authorization: `Bearer ${accessToken}`,
+        'Content-Type': 'application/json',
+        'anthropic-version': '2023-06-01',
+        'anthropic-beta': TRIGGERS_BETA,
+        'x-organization-uuid': orgUUID,
+      }

-    const { action, trigger_id, body } = input
-    let method: 'GET' | 'POST'
-    let url: string
-    let data: unknown
-    switch (action) {
-      case 'list':
-        method = 'GET'
-        url = base
-        break
-      case 'get':
-        if (!trigger_id) throw new Error('get requires trigger_id')
-        method = 'GET'
-        url = `${base}/${trigger_id}`
-        break
-      case 'create':
-        if (!body) throw new Error('create requires body')
-        method = 'POST'
-        url = base
-        data = body
-        break
-      case 'update':
-        if (!trigger_id) throw new Error('update requires trigger_id')
-        if (!body) throw new Error('update requires body')
-        method = 'POST'
-        url = `${base}/${trigger_id}`
-        data = body
-        break
-      case 'run':
-        if (!trigger_id) throw new Error('run requires trigger_id')
-        method = 'POST'
-        url = `${base}/${trigger_id}/run`
-        data = {}
-        break
-    }
+      const { action, trigger_id, body } = input
+      let method: 'GET' | 'POST'
+      let url: string
+      let data: unknown
+      switch (action) {
+        case 'list':
+          method = 'GET'
+          url = base
+          break
+        case 'get':
+          if (!trigger_id) throw new Error('get requires trigger_id')
+          method = 'GET'
+          url = `${base}/${trigger_id}`
+          break
+        case 'create':
+          if (!body) throw new Error('create requires body')
+          method = 'POST'
+          url = base
+          data = body
+          break
+        case 'update':
+          if (!trigger_id) throw new Error('update requires trigger_id')
+          if (!body) throw new Error('update requires body')
+          method = 'POST'
+          url = `${base}/${trigger_id}`
+          data = body
+          break
+        case 'run':
+          if (!trigger_id) throw new Error('run requires trigger_id')
+          method = 'POST'
+          url = `${base}/${trigger_id}/run`
+          data = {}
+          break
+      }

-    const res = await axios.request({
-      method,
-      url,
-      headers,
-      data,
-      timeout: 20_000,
-      signal: context.abortController.signal,
-      validateStatus: () => true,
-    })
-
-    return {
-      data: {
+      const res = await axios.request({
+        method,
+        url,
+        headers,
+        data,
+        timeout: 20_000,
+        signal: context.abortController.signal,
+        validateStatus: () => true,
+      })
+      const audit = await appendRemoteTriggerAuditRecord({
+        ...auditBase,
+        ok: res.status >= 200 && res.status < 300,
        status: res.status,
-        json: jsonStringify(res.data),
-      },
+      })
+
+      return {
+        data: {
+          status: res.status,
+          json: jsonStringify(res.data),
+          audit_id: audit.auditId,
+        },
+      }
+    } catch (error) {
+      await appendRemoteTriggerAuditRecord({
+        ...auditBase,
+        ok: false,
+        error: error instanceof Error ? error.message : String(error),
+      })
+      throw error
    }
  },
  mapToolResultToToolResultBlockParam(output, toolUseID) {
--- a/packages/builtin-tools/src/tools/RemoteTriggerTool/tests/RemoteTriggerTool.test.ts
+++ b/packages/builtin-tools/src/tools/RemoteTriggerTool/tests/RemoteTriggerTool.test.ts
@@ -0,0 +1,91 @@
+import { afterEach, beforeEach, describe, expect, mock, test } from 'bun:test'
+import { mkdir, readFile, rm } from 'fs/promises'
+import { tmpdir } from 'os'
+import { join } from 'path'
+import {
+  resetStateForTests,
+  setOriginalCwd,
+  setProjectRoot,
+} from 'src/bootstrap/state.js'
+
+let requestStatus = 200
+
+mock.module('axios', () => ({
+  default: {
+    request: async () => ({
+      status: requestStatus,
+      data: { ok: requestStatus >= 200 && requestStatus < 300 },
+    }),
+  },
+}))
+
+mock.module('src/utils/auth.js', () => ({
+  checkAndRefreshOAuthTokenIfNeeded: async () => {},
+  getClaudeAIOAuthTokens: () => ({ accessToken: 'token' }),
+}))
+
+mock.module('src/services/oauth/client.js', () => ({
+  getOrganizationUUID: async () => 'org',
+}))
+
+mock.module('src/constants/oauth.js', () => ({
+  getOauthConfig: () => ({ BASE_API_URL: 'https://example.test' }),
+}))
+
+let cwd = ''
+let previousCwd = ''
+
+beforeEach(async () => {
+  requestStatus = 200
+  previousCwd = process.cwd()
+  cwd = join(tmpdir(), `remote-trigger-tool-${Date.now()}-${Math.random().toString(16).slice(2)}`)
+  await mkdir(cwd, { recursive: true })
+  process.chdir(cwd)
+  resetStateForTests()
+  setOriginalCwd(cwd)
+  setProjectRoot(cwd)
+})
+
+afterEach(async () => {
+  resetStateForTests()
+  process.chdir(previousCwd)
+  await rm(cwd, { recursive: true, force: true })
+})
+
+describe('RemoteTriggerTool audit', () => {
+  test('writes an audit record for successful remote calls', async () => {
+    const { RemoteTriggerTool } = await import('../RemoteTriggerTool')
+    const result = await RemoteTriggerTool.call(
+      { action: 'run', trigger_id: 'trigger-1' },
+      { abortController: new AbortController() } as any,
+    )
+
+    expect(result.data.audit_id).toBeString()
+    const raw = await readFile(
+      join(cwd, '.claude', 'remote-trigger-audit.jsonl'),
+      'utf-8',
+    )
+    expect(raw).toContain('"action":"run"')
+    expect(raw).toContain('"triggerId":"trigger-1"')
+    expect(raw).toContain('"ok":true')
+  })
+
+  test('writes an audit record before rethrowing validation failures', async () => {
+    const { RemoteTriggerTool } = await import('../RemoteTriggerTool')
+
+    await expect(
+      RemoteTriggerTool.call(
+        { action: 'run' },
+        { abortController: new AbortController() } as any,
+      ),
+    ).rejects.toThrow('run requires trigger_id')
+
+    const raw = await readFile(
+      join(cwd, '.claude', 'remote-trigger-audit.jsonl'),
+      'utf-8',
+    )
+    expect(raw).toContain('"action":"run"')
+    expect(raw).toContain('"ok":false')
+    expect(raw).toContain('run requires trigger_id')
+  })
+})
--- a/packages/builtin-tools/src/tools/TeamDeleteTool/TeamDeleteTool.ts
+++ b/packages/builtin-tools/src/tools/TeamDeleteTool/TeamDeleteTool.ts
@@ -14,11 +14,26 @@ import {
 } from 'src/utils/swarm/teamHelpers.js'
 import { clearTeammateColors } from 'src/utils/swarm/teammateLayoutManager.js'
 import { clearLeaderTeamName } from 'src/utils/tasks.js'
+import { ensureBackendsRegistered, getBackendByType, getInProcessBackend } from 'src/utils/swarm/backends/registry.js'
+import { createPaneBackendExecutor } from 'src/utils/swarm/backends/PaneBackendExecutor.js'
+import { isPaneBackend } from 'src/utils/swarm/backends/types.js'
+import { sleep } from 'src/utils/sleep.js'
 import { TEAM_DELETE_TOOL_NAME } from './constants.js'
 import { getPrompt } from './prompt.js'
 import { renderToolResultMessage, renderToolUseMessage } from './UI.js'

-const inputSchema = lazySchema(() => z.strictObject({}))
+const inputSchema = lazySchema(() =>
+  z.strictObject({
+    wait_ms: z
+      .number()
+      .min(0)
+      .max(30_000)
+      .optional()
+      .describe(
+        'Optional time to wait for active teammates to acknowledge shutdown before cleanup.',
+      ),
+  }),
+)
 type InputSchema = ReturnType<typeof inputSchema>

 export type Output = {
@@ -68,7 +83,7 @@ export const TeamDeleteTool: Tool<InputSchema, Output> = buildTool({
    }
  },

-  async call(_input, context) {
+  async call(input, context) {
    const { setAppState, getAppState } = context
    const appState = getAppState()
    const teamName = appState.teamContext?.teamName
@@ -87,13 +102,82 @@ export const TeamDeleteTool: Tool<InputSchema, Output> = buildTool({
        const activeMembers = nonLeadMembers.filter(m => m.isActive !== false)

        if (activeMembers.length > 0) {
-          const memberNames = activeMembers.map(m => m.name).join(', ')
-          return {
-            data: {
-              success: false,
-              message: `Cannot cleanup team with ${activeMembers.length} active member(s): ${memberNames}. Use requestShutdown to gracefully terminate teammates first.`,
-              team_name: teamName,
-            },
+          const requested: string[] = []
+          for (const member of activeMembers) {
+            let sent = false
+            if (member.backendType === 'in-process') {
+              const executor = getInProcessBackend()
+              executor.setContext?.(context)
+              sent = await executor.terminate(
+                member.agentId,
+                'Team cleanup requested by team lead',
+              )
+            } else if (member.backendType && isPaneBackend(member.backendType)) {
+              await ensureBackendsRegistered()
+              const executor = createPaneBackendExecutor(
+                getBackendByType(member.backendType),
+              )
+              executor.setContext?.(context)
+              sent = await executor.terminate(
+                member.agentId,
+                'Team cleanup requested by team lead',
+              )
+            }
+            if (sent) {
+              requested.push(member.name)
+            }
+          }
+          const waitMs = input.wait_ms ?? 0
+          if (waitMs > 0 && requested.length > 0) {
+            const deadline = Date.now() + waitMs
+            while (Date.now() < deadline) {
+              await sleep(Math.min(250, Math.max(0, deadline - Date.now())))
+              const refreshed = readTeamFile(teamName)
+              const stillActive =
+                refreshed?.members.filter(
+                  m => m.name !== TEAM_LEAD_NAME && m.isActive !== false,
+                ) ?? []
+              if (stillActive.length === 0) {
+                break
+              }
+            }
+            const refreshed = readTeamFile(teamName)
+            const stillActive =
+              refreshed?.members.filter(
+                m => m.name !== TEAM_LEAD_NAME && m.isActive !== false,
+              ) ?? []
+            if (stillActive.length === 0) {
+              // Fall through to cleanup with the refreshed team file state.
+            } else {
+              const memberNames = stillActive.map(m => m.name).join(', ')
+              return {
+                data: {
+                  success: false,
+                  message: `Shutdown requested for active teammate(s): ${requested.join(', ')}. Cleanup is still blocked after waiting ${waitMs}ms: ${memberNames}.`,
+                  team_name: teamName,
+                },
+              }
+            }
+          }
+          const latestTeamFile = readTeamFile(teamName)
+          const latestActiveMembers =
+            latestTeamFile?.members.filter(
+              m => m.name !== TEAM_LEAD_NAME && m.isActive !== false,
+            ) ?? []
+          if (latestActiveMembers.length === 0) {
+            // Continue to cleanup below.
+          } else {
+            const memberNames = latestActiveMembers.map(m => m.name).join(', ')
+            return {
+              data: {
+                success: false,
+                message:
+                  requested.length > 0
+                    ? `Shutdown requested for active teammate(s): ${requested.join(', ')}. Cleanup is blocked until they exit: ${memberNames}.`
+                    : `Cannot cleanup team with ${latestActiveMembers.length} active member(s): ${memberNames}. Use requestShutdown to gracefully terminate teammates first.`,
+                team_name: teamName,
+              },
+            }
          }
        }
      }
--- a/packages/builtin-tools/src/tools/WebBrowserTool/WebBrowserTool.ts
+++ b/packages/builtin-tools/src/tools/WebBrowserTool/WebBrowserTool.ts
@@ -9,19 +9,11 @@ const inputSchema = lazySchema(() =>
  z.strictObject({
    url: z
      .string()
-      .describe('URL to navigate to in the browser.'),
+      .describe('URL to fetch and extract content from.'),
    action: z
-      .enum(['navigate', 'screenshot', 'click', 'type', 'scroll'])
+      .enum(['navigate', 'screenshot'])
      .optional()
-      .describe('Browser action to perform. Defaults to "navigate".'),
-    selector: z
-      .string()
-      .optional()
-      .describe('CSS selector for click/type actions.'),
-    text: z
-      .string()
-      .optional()
-      .describe('Text to type when action is "type".'),
+      .describe('Action to perform. "navigate" fetches page content (default). "screenshot" returns a text snapshot of the page.'),
  }),
 )
 type InputSchema = ReturnType<typeof inputSchema>
@@ -45,16 +37,24 @@ export const WebBrowserTool = buildTool({
  },

  async description() {
-    return 'Browse the web using an embedded browser'
+    return 'Fetch and read web page content via HTTP'
  },
  async prompt() {
-    return `Open and interact with web pages in an embedded browser. Supports navigation, screenshots, clicking, typing, and scrolling.
+    return `Fetch web pages via HTTP and extract their text content. This is a lightweight browser tool (HTTP fetch, not a full browser engine).
+
+Supported actions:
+- navigate: Fetch a URL and extract page title + text content
+- screenshot: Same as navigate (returns text snapshot, not a visual screenshot)
+
+Limitations:
+- No JavaScript execution — only sees server-rendered HTML
+- click/type/scroll require a full browser runtime (not available)
+- For full browser interaction, use the Claude-in-Chrome MCP tools instead

 Use this for:
- Viewing web pages and their content
- Taking screenshots of UI
- Interacting with web applications
- Testing web endpoints with full browser rendering`
+- Reading web page content and documentation
+- Checking API endpoints that return HTML
+- Quick page title/content extraction`
  },

  isConcurrencySafe() {
@@ -85,12 +85,84 @@ Use this for:
  },

  async call(input: BrowserInput) {
-    // Browser integration requires the WEB_BROWSER_TOOL runtime (Bun WebView).
+    const action = input.action ?? 'navigate'
+
+    if (action === 'navigate' || action === 'screenshot') {
+      // Fetch the page content via HTTP
+      try {
+        const response = await fetch(input.url, {
+          headers: {
+            'User-Agent':
+              'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0 Safari/537.36',
+            Accept:
+              'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
+          },
+          redirect: 'follow',
+        })
+
+        if (!response.ok) {
+          return {
+            data: {
+              title: `HTTP ${response.status}`,
+              url: input.url,
+              content: `Error: ${response.status} ${response.statusText}`,
+            },
+          }
+        }
+
+        const html = await response.text()
+
+        // Extract title
+        const titleMatch = html.match(/<title[^>]*>([^<]*)<\/title>/i)
+        const title = titleMatch?.[1]?.trim() ?? ''
+
+        // Extract text content (strip HTML tags, scripts, styles)
+        let textContent = html
+          .replace(/<script[\s\S]*?<\/script>/gi, '')
+          .replace(/<style[\s\S]*?<\/style>/gi, '')
+          .replace(/<[^>]+>/g, ' ')
+          .replace(/\s+/g, ' ')
+          .trim()
+
+        // Truncate to reasonable size
+        if (textContent.length > 50_000) {
+          textContent = textContent.slice(0, 50_000) + '\n[truncated]'
+        }
+
+        if (action === 'screenshot') {
+          return {
+            data: {
+              title,
+              url: response.url,
+              content: `[Text snapshot — visual screenshots require Chrome browser tools]\n\n${textContent}`,
+            },
+          }
+        }
+
+        return {
+          data: {
+            title,
+            url: response.url,
+            content: textContent,
+          },
+        }
+      } catch (err) {
+        return {
+          data: {
+            title: 'Error',
+            url: input.url,
+            content: `Failed to fetch: ${err instanceof Error ? err.message : String(err)}`,
+          },
+        }
+      }
+    }
+
+    // Unreachable — schema only allows navigate/screenshot
    return {
      data: {
        title: '',
        url: input.url,
-        content: 'Web browser requires the WEB_BROWSER_TOOL runtime.',
+        content: `Unknown action "${action}".`,
      },
    }
  },
--- a/packages/builtin-tools/src/tools/WebBrowserTool/tests/WebBrowserTool.test.ts
+++ b/packages/builtin-tools/src/tools/WebBrowserTool/tests/WebBrowserTool.test.ts
@@ -0,0 +1,94 @@
+import { describe, test, expect, beforeAll, afterAll } from 'bun:test'
+
+// Mock fetch directly — avoids flaky dependency on external hosts AND
+// pollution by other tests that call setGlobalDispatcher (proxy agents make
+// localhost fetches return 500 in the full-suite run).
+const realFetch = globalThis.fetch
+
+beforeAll(() => {
+  globalThis.fetch = (async (
+    input: string | URL | Request,
+    _init?: RequestInit,
+  ) => {
+    const url = typeof input === 'string' ? input : input.toString()
+    if (url === 'not-a-url' || !url.startsWith('http')) {
+      throw new TypeError('Failed to fetch')
+    }
+    const body =
+      '<!doctype html><html><head><title>Example Domain</title></head>' +
+      '<body><h1>Example Domain</h1><p>Sample content.</p></body></html>'
+    const res = new Response(body, {
+      status: 200,
+      headers: { 'content-type': 'text/html' },
+    })
+    // Make response.url match the request URL so tests can assert on it.
+    Object.defineProperty(res, 'url', { value: url, configurable: true })
+    return res
+  }) as typeof fetch
+})
+
+afterAll(() => {
+  globalThis.fetch = realFetch
+})
+
+describe('WebBrowserTool', () => {
+  test('tool exports and metadata', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    expect(WebBrowserTool).toBeDefined()
+    expect(WebBrowserTool.name).toBe('WebBrowser')
+    expect(typeof WebBrowserTool.call).toBe('function')
+    expect(WebBrowserTool.userFacingName()).toBe('Browser')
+    expect(WebBrowserTool.isReadOnly()).toBe(true)
+  })
+
+  test('description reflects browser-lite', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const desc = await WebBrowserTool.description()
+    expect(desc).toContain('HTTP')
+    expect(desc).not.toContain('embedded browser')
+  })
+
+  test('prompt mentions limitations', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const prompt = await WebBrowserTool.prompt()
+    expect(prompt).toContain('Limitations')
+    expect(prompt).toContain('No JavaScript')
+    expect(prompt).toContain('Claude-in-Chrome')
+  })
+
+  test('navigate fetches URL', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const result = await WebBrowserTool.call({
+      url: 'https://example.com',
+    } as any)
+    expect(result.data.title).toBe('Example Domain')
+    expect(result.data.url).toContain('example.com')
+    expect(result.data.content).toContain('Example Domain')
+  }, 15000)
+
+  test('screenshot returns text snapshot', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const result = await WebBrowserTool.call({
+      url: 'https://example.com',
+      action: 'screenshot',
+    } as any)
+    expect(result.data.content).toContain('Text snapshot')
+    expect(result.data.content).toContain('Example Domain')
+  }, 15000)
+
+  test('schema only allows navigate and screenshot', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const schema = WebBrowserTool.inputSchema
+    const parseResult = schema.safeParse({
+      url: 'https://example.com',
+      action: 'click',
+    })
+    expect(parseResult.success).toBe(false)
+  })
+
+  test('invalid URL returns error', async () => {
+    const { WebBrowserTool } = await import('../WebBrowserTool.js')
+    const result = await WebBrowserTool.call({ url: 'not-a-url' } as any)
+    expect(result.data.content).toContain('Failed to fetch')
+  })
+})
--- a/packages/builtin-tools/src/tools/WebSearchTool/adapters/index.ts
+++ b/packages/builtin-tools/src/tools/WebSearchTool/adapters/index.ts
@@ -17,17 +17,37 @@ export type {
  WebSearchAdapter,
 } from './types.js'

+/**
+ * Check if the current session uses a third-party (non-Anthropic) API provider.
+ * These providers don't support Anthropic's server_tools (server-side web search),
+ * so they must fall back to the Bing scraper adapter.
+ */
+function isThirdPartyProvider(): boolean {
+  return !!(
+    process.env.CLAUDE_CODE_USE_OPENAI ||
+    process.env.CLAUDE_CODE_USE_GEMINI ||
+    process.env.CLAUDE_CODE_USE_GROK
+  )
+}
+
 let cachedAdapter: WebSearchAdapter | null = null
 let cachedAdapterKey: 'api' | 'bing' | 'brave' | 'exa' | null = null

 export function createAdapter(): WebSearchAdapter {
  const envAdapter = process.env.WEB_SEARCH_ADAPTER
+  // Priority:
+  //   1. Explicit env override (WEB_SEARCH_ADAPTER=api|bing|brave)
+  //   2. Third-party provider (OpenAI/Gemini/Grok) → bing (no server_tools support)
+  //   3. First-party Anthropic API → api (server-side web search + connector_text)
+  //   4. Fallback → bing
  const adapterKey =
    envAdapter === 'api' || envAdapter === 'bing' || envAdapter === 'brave' || envAdapter === 'exa'
      ? envAdapter
-      : isFirstPartyAnthropicBaseUrl()
-        ? 'api'
-        : 'exa'
+      : isThirdPartyProvider()
+        ? 'bing'
+        : isFirstPartyAnthropicBaseUrl()
+          ? 'api'
+          : 'bing'

  if (cachedAdapter && cachedAdapterKey === adapterKey) return cachedAdapter

--- a/packages/builtin-tools/src/tools/WorkflowTool/WorkflowTool.ts
+++ b/packages/builtin-tools/src/tools/WorkflowTool/WorkflowTool.ts
@@ -1,18 +1,358 @@
+import { randomUUID } from 'crypto'
+import { mkdir, readdir, readFile, writeFile } from 'fs/promises'
+import { join, parse } from 'path'
 import { z } from 'zod/v4'
 import type { ToolResultBlockParam } from 'src/Tool.js'
 import { buildTool } from 'src/Tool.js'
 import { truncate } from 'src/utils/format.js'
-import { WORKFLOW_TOOL_NAME } from './constants.js'
+import { safeParseJSON } from 'src/utils/json.js'
+import {
+  WORKFLOW_DIR_NAME,
+  WORKFLOW_FILE_EXTENSIONS,
+  WORKFLOW_TOOL_NAME,
+} from './constants.js'
+
+const WORKFLOW_RUNS_DIR = '.claude/workflow-runs'

 const inputSchema = z.object({
  workflow: z.string().describe('Name of the workflow to execute'),
  args: z.string().optional().describe('Arguments to pass to the workflow'),
+  action: z
+    .enum(['start', 'status', 'advance', 'cancel', 'list'])
+    .optional()
+    .describe('Workflow action. Defaults to start.'),
+  run_id: z
+    .string()
+    .optional()
+    .describe('Workflow run id for status, advance, or cancel.'),
 })
 type Input = typeof inputSchema
 type WorkflowInput = z.infer<Input>

+type WorkflowStepStatus = 'pending' | 'running' | 'completed' | 'cancelled'
+
+type WorkflowStep = {
+  name: string
+  prompt: string
+  status: WorkflowStepStatus
+  startedAt?: number
+  completedAt?: number
+}
+
+type WorkflowRun = {
+  runId: string
+  workflow: string
+  args?: string
+  status: 'running' | 'completed' | 'cancelled'
+  createdAt: number
+  updatedAt: number
+  currentStepIndex: number
+  steps: WorkflowStep[]
+}
+
 type WorkflowOutput = { output: string }

+async function findWorkflowFile(
+  workflowDir: string,
+  workflow: string,
+): Promise<{ path: string; content: string } | null> {
+  for (const ext of WORKFLOW_FILE_EXTENSIONS) {
+    const path = join(workflowDir, `${workflow}${ext}`)
+    try {
+      return { path, content: await readFile(path, 'utf-8') }
+    } catch {
+      // try next
+    }
+  }
+  return null
+}
+
+async function listAvailableWorkflows(workflowDir: string): Promise<string[]> {
+  try {
+    const files = await readdir(workflowDir)
+    return files
+      .filter(f => WORKFLOW_FILE_EXTENSIONS.includes(parse(f).ext.toLowerCase()))
+      .map(f => parse(f).name)
+      .sort()
+  } catch {
+    return []
+  }
+}
+
+function workflowRunPath(cwd: string, runId: string): string {
+  return join(cwd, WORKFLOW_RUNS_DIR, `${runId}.json`)
+}
+
+async function readWorkflowRun(
+  cwd: string,
+  runId: string,
+): Promise<WorkflowRun | null> {
+  try {
+    const parsed = safeParseJSON(
+      await readFile(workflowRunPath(cwd, runId), 'utf-8'),
+      false,
+    ) as Partial<WorkflowRun> | null
+    if (
+      !parsed ||
+      typeof parsed.runId !== 'string' ||
+      typeof parsed.workflow !== 'string' ||
+      !Array.isArray(parsed.steps)
+    ) {
+      return null
+    }
+    return parsed as WorkflowRun
+  } catch {
+    return null
+  }
+}
+
+async function writeWorkflowRun(cwd: string, run: WorkflowRun): Promise<void> {
+  await mkdir(join(cwd, WORKFLOW_RUNS_DIR), { recursive: true })
+  await writeFile(
+    workflowRunPath(cwd, run.runId),
+    JSON.stringify(run, null, 2) + '\n',
+    'utf-8',
+  )
+}
+
+async function listWorkflowRuns(cwd: string): Promise<WorkflowRun[]> {
+  let files: string[]
+  try {
+    files = await readdir(join(cwd, WORKFLOW_RUNS_DIR))
+  } catch {
+    return []
+  }
+  const runs = await Promise.all(
+    files
+      .filter(f => f.endsWith('.json'))
+      .map(f => readWorkflowRun(cwd, f.slice(0, -'.json'.length))),
+  )
+  return runs
+    .filter((run): run is WorkflowRun => run !== null)
+    .sort((a, b) => b.updatedAt - a.updatedAt)
+}
+
+function parseMarkdownSteps(content: string): WorkflowStep[] {
+  const steps: WorkflowStep[] = []
+  for (const rawLine of content.split('\n')) {
+    const line = rawLine.trim()
+    const taskMatch = line.match(/^[-*]\s+\[[ xX]\]\s+(.+)$/)
+    const bulletMatch = line.match(/^[-*]\s+(.+)$/)
+    const numberedMatch = line.match(/^\d+[.)]\s+(.+)$/)
+    const text = taskMatch?.[1] ?? bulletMatch?.[1] ?? numberedMatch?.[1]
+    if (!text) continue
+    steps.push({ name: text.slice(0, 80), prompt: text, status: 'pending' })
+  }
+  return steps
+}
+
+function parseYamlSteps(content: string): WorkflowStep[] {
+  const steps: WorkflowStep[] = []
+  let current: Partial<WorkflowStep> | null = null
+  const flush = () => {
+    if (!current) return
+    const prompt = current.prompt ?? current.name
+    if (current.name && prompt) {
+      steps.push({
+        name: current.name,
+        prompt,
+        status: 'pending',
+      })
+    }
+    current = null
+  }
+
+  for (const rawLine of content.split('\n')) {
+    const line = rawLine.trim()
+    const stepText = line.match(/^-\s+(.+)$/)?.[1]
+    if (stepText) {
+      flush()
+      const inlineName = stepText.match(/^name:\s*(.+)$/)?.[1]
+      current = {
+        name: inlineName ?? stepText,
+        prompt: inlineName ? undefined : stepText,
+      }
+      continue
+    }
+    const name = line.match(/^name:\s*(.+)$/)?.[1]
+    if (name) {
+      if (!current) current = {}
+      current.name = name
+      continue
+    }
+    const prompt = line.match(/^(prompt|run|command):\s*(.+)$/)?.[2]
+    if (prompt) {
+      if (!current) current = {}
+      current.prompt = prompt
+    }
+  }
+  flush()
+  return steps
+}
+
+function parseWorkflowSteps(filePath: string, content: string): WorkflowStep[] {
+  const ext = parse(filePath).ext.toLowerCase()
+  const steps =
+    ext === '.md' ? parseMarkdownSteps(content) : parseYamlSteps(content)
+  if (steps.length > 0) {
+    return steps
+  }
+  return [
+    {
+      name: 'Execute workflow',
+      prompt: content.trim(),
+      status: 'pending',
+    },
+  ]
+}
+
+function formatStep(step: WorkflowStep, index: number): string {
+  return `Step ${index + 1}: ${step.name}\n${step.prompt}`
+}
+
+function formatRunStatus(run: WorkflowRun): string {
+  const lines = [
+    `Workflow run: ${run.runId}`,
+    `Workflow: ${run.workflow}`,
+    `Status: ${run.status}`,
+    `Current step: ${run.steps[run.currentStepIndex]?.name ?? 'none'}`,
+    `Steps: ${run.steps.length}`,
+  ]
+  for (let i = 0; i < run.steps.length; i += 1) {
+    const step = run.steps[i]!
+    lines.push(`  ${i + 1}. [${step.status}] ${step.name}`)
+  }
+  return lines.join('\n')
+}
+
+async function startWorkflow(
+  input: WorkflowInput,
+  cwd: string,
+): Promise<WorkflowOutput> {
+  const workflowDir = join(cwd, WORKFLOW_DIR_NAME)
+  const found = await findWorkflowFile(workflowDir, input.workflow)
+  if (!found) {
+    const available = await listAvailableWorkflows(workflowDir)
+    const hint =
+      available.length > 0
+        ? `\nAvailable workflows: ${available.join(', ')}`
+        : `\nNo workflows found in ${WORKFLOW_DIR_NAME}/. Create .md or .yaml files there.`
+    return { output: `Error: Workflow "${input.workflow}" not found.${hint}` }
+  }
+
+  const steps = parseWorkflowSteps(found.path, found.content)
+  const now = Date.now()
+  steps[0] = { ...steps[0]!, status: 'running', startedAt: now }
+  const run: WorkflowRun = {
+    runId: randomUUID(),
+    workflow: input.workflow,
+    ...(input.args ? { args: input.args } : {}),
+    status: 'running',
+    createdAt: now,
+    updatedAt: now,
+    currentStepIndex: 0,
+    steps,
+  }
+  await writeWorkflowRun(cwd, run)
+
+  const argsSection = input.args ? `\n\nArguments:\n${input.args}` : ''
+  return {
+    output: [
+      `Workflow run started`,
+      `run_id: ${run.runId}`,
+      `workflow: ${run.workflow}`,
+      '',
+      formatStep(steps[0]!, 0),
+      argsSection,
+      '',
+      `When this step is complete, call Workflow with action="advance" and run_id="${run.runId}".`,
+    ].join('\n'),
+  }
+}
+
+async function getRunOrError(
+  cwd: string,
+  runId: string | undefined,
+): Promise<{ run?: WorkflowRun; output?: string }> {
+  if (!runId) return { output: 'Error: run_id is required for this action.' }
+  const run = await readWorkflowRun(cwd, runId)
+  if (!run) return { output: `Error: Workflow run "${runId}" not found.` }
+  return { run }
+}
+
+async function advanceWorkflow(
+  cwd: string,
+  runId: string | undefined,
+): Promise<WorkflowOutput> {
+  const found = await getRunOrError(cwd, runId)
+  if (!found.run) return { output: found.output! }
+  const run = found.run
+  const now = Date.now()
+  const current = run.steps[run.currentStepIndex]
+  if (current && current.status === 'running') {
+    current.status = 'completed'
+    current.completedAt = now
+  }
+  const nextIndex = run.currentStepIndex + 1
+  if (nextIndex >= run.steps.length) {
+    run.status = 'completed'
+    run.updatedAt = now
+    await writeWorkflowRun(cwd, run)
+    return { output: `Workflow completed\nrun_id: ${run.runId}` }
+  }
+  run.currentStepIndex = nextIndex
+  run.steps[nextIndex] = {
+    ...run.steps[nextIndex]!,
+    status: 'running',
+    startedAt: now,
+  }
+  run.updatedAt = now
+  await writeWorkflowRun(cwd, run)
+  return {
+    output: [
+      `Next workflow step`,
+      `run_id: ${run.runId}`,
+      '',
+      formatStep(run.steps[nextIndex]!, nextIndex),
+      '',
+      `When this step is complete, call Workflow with action="advance" and run_id="${run.runId}".`,
+    ].join('\n'),
+  }
+}
+
+async function cancelWorkflow(
+  cwd: string,
+  runId: string | undefined,
+): Promise<WorkflowOutput> {
+  const found = await getRunOrError(cwd, runId)
+  if (!found.run) return { output: found.output! }
+  const run = found.run
+  const now = Date.now()
+  run.status = 'cancelled'
+  run.updatedAt = now
+  for (const step of run.steps) {
+    if (step.status === 'pending' || step.status === 'running') {
+      step.status = 'cancelled'
+    }
+  }
+  await writeWorkflowRun(cwd, run)
+  return { output: `Workflow cancelled\nrun_id: ${run.runId}` }
+}
+
+async function listWorkflowRunsForOutput(cwd: string): Promise<WorkflowOutput> {
+  const runs = await listWorkflowRuns(cwd)
+  if (runs.length === 0) return { output: 'No workflow runs recorded.' }
+  return {
+    output: runs
+      .slice(0, 20)
+      .map(
+        run =>
+          `${run.runId} | ${run.workflow} | ${run.status} | step=${run.steps[run.currentStepIndex]?.name ?? 'none'} | updated=${new Date(run.updatedAt).toLocaleString()}`,
+      )
+      .join('\n'),
+  }
+}
+
 export const WorkflowTool = buildTool({
  name: WORKFLOW_TOOL_NAME,
  searchHint: 'execute user-defined workflow scripts',
@@ -22,21 +362,25 @@ export const WorkflowTool = buildTool({
  inputSchema,

  async description() {
-    return 'Execute a user-defined workflow script from .claude/workflows/'
+    return 'Execute and track a user-defined workflow from .claude/workflows/'
  },
  async prompt() {
-    return `Use the Workflow tool to execute user-defined workflow scripts located in .claude/workflows/. Workflows are YAML or Markdown files that define a sequence of steps for common development tasks.
+    return `Use the Workflow tool to run user-defined workflows located in .claude/workflows/. Workflows may be Markdown checklists/lists or YAML files with steps.

-Guidelines:
- Specify the workflow name to execute (must match a file in .claude/workflows/)
- Optionally pass arguments that the workflow can use
- Workflows run in the context of the current project`
+Actions:
+- start (default): create a persisted workflow run and return the first step to execute
+- advance: mark the current step complete and return the next step
+- status: inspect a workflow run by run_id
+- cancel: cancel a workflow run
+- list: list recent workflow runs
+
+Workflow run state is persisted in .claude/workflow-runs/.`
  },
  userFacingName() {
    return 'Workflow'
  },
-  isReadOnly() {
-    return false
+  isReadOnly(input) {
+    return input.action === 'status' || input.action === 'list'
  },
  isEnabled() {
    return true
@@ -44,10 +388,10 @@ Guidelines:

  renderToolUseMessage(input: Partial<WorkflowInput>) {
    const name = input.workflow ?? 'unknown'
-    if (input.args) {
-      return `Workflow: ${name} ${input.args}`
-    }
-    return `Workflow: ${name}`
+    const action = input.action ?? 'start'
+    return input.args
+      ? `Workflow: ${action} ${name} ${input.args}`
+      : `Workflow: ${action} ${name}`
  },

  mapToolResultToToolResultBlockParam(
@@ -61,14 +405,26 @@ Guidelines:
    }
  },

-  async call(_input: WorkflowInput, _context, _progress) {
-    // Workflow execution is wired by the WORKFLOW_SCRIPTS feature bootstrap.
-    // Without it, this tool is not functional.
-    return {
-      data: {
-        output:
-          'Error: Workflow execution requires the WORKFLOW_SCRIPTS runtime.',
-      },
+  async call(input: WorkflowInput) {
+    const cwd = process.cwd()
+    const action = input.action ?? 'start'
+    switch (action) {
+      case 'start':
+        return { data: await startWorkflow(input, cwd) }
+      case 'status': {
+        const found = await getRunOrError(cwd, input.run_id)
+        return {
+          data: {
+            output: found.run ? formatRunStatus(found.run) : found.output!,
+          },
+        }
+      }
+      case 'advance':
+        return { data: await advanceWorkflow(cwd, input.run_id) }
+      case 'cancel':
+        return { data: await cancelWorkflow(cwd, input.run_id) }
+      case 'list':
+        return { data: await listWorkflowRunsForOutput(cwd) }
    }
  },
 })
--- a/packages/builtin-tools/src/tools/WorkflowTool/tests/WorkflowTool.test.ts
+++ b/packages/builtin-tools/src/tools/WorkflowTool/tests/WorkflowTool.test.ts
@@ -0,0 +1,99 @@
+import { afterEach, beforeEach, describe, expect, test } from 'bun:test'
+import { mkdir, readFile, rm, writeFile } from 'node:fs/promises'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+import { WorkflowTool } from '../WorkflowTool'
+
+let cwd: string
+let previousCwd: string
+
+beforeEach(async () => {
+  previousCwd = process.cwd()
+  cwd = join(tmpdir(), `workflow-tool-${Date.now()}-${Math.random().toString(16).slice(2)}`)
+  await mkdir(join(cwd, '.claude', 'workflows'), { recursive: true })
+  process.chdir(cwd)
+})
+
+afterEach(async () => {
+  process.chdir(previousCwd)
+  await rm(cwd, { recursive: true, force: true })
+})
+
+describe('WorkflowTool', () => {
+  test('starts a workflow run and persists step state', async () => {
+    await writeFile(
+      join(cwd, '.claude', 'workflows', 'release.md'),
+      [
+        '# Release',
+        '',
+        '- [ ] Run tests',
+        '- [ ] Build package',
+      ].join('\n'),
+    )
+
+    const result = await WorkflowTool.call({ workflow: 'release' })
+
+    expect(result.data.output).toContain('Workflow run started')
+    expect(result.data.output).toContain('Run tests')
+    const match = result.data.output.match(/run_id: ([a-f0-9-]+)/)
+    expect(match?.[1]).toBeString()
+
+    const raw = await readFile(
+      join(cwd, '.claude', 'workflow-runs', `${match![1]}.json`),
+      'utf-8',
+    )
+    const run = JSON.parse(raw)
+    expect(run.workflow).toBe('release')
+    expect(run.status).toBe('running')
+    expect(run.steps).toHaveLength(2)
+    expect(run.steps[0].status).toBe('running')
+    expect(run.steps[1].status).toBe('pending')
+  })
+
+  test('advances a workflow run through completion', async () => {
+    await writeFile(
+      join(cwd, '.claude', 'workflows', 'audit.yaml'),
+      [
+        'steps:',
+        '  - name: Inspect',
+        '    prompt: Inspect the code',
+        '  - name: Verify',
+        '    prompt: Run focused tests',
+      ].join('\n'),
+    )
+
+    const started = await WorkflowTool.call({ workflow: 'audit' })
+    const runId = started.data.output.match(/run_id: ([a-f0-9-]+)/)![1]!
+
+    const next = await WorkflowTool.call(
+      { workflow: 'audit', action: 'advance', run_id: runId },
+    )
+    expect(next.data.output).toContain('Next workflow step')
+    expect(next.data.output).toContain('Run focused tests')
+
+    const done = await WorkflowTool.call(
+      { workflow: 'audit', action: 'advance', run_id: runId },
+    )
+    expect(done.data.output).toContain('Workflow completed')
+  })
+
+  test('lists and cancels workflow runs', async () => {
+    await writeFile(
+      join(cwd, '.claude', 'workflows', 'cleanup.md'),
+      '- Remove stale files',
+    )
+
+    const started = await WorkflowTool.call({ workflow: 'cleanup' })
+    const runId = started.data.output.match(/run_id: ([a-f0-9-]+)/)![1]!
+
+    const listed = await WorkflowTool.call(
+      { workflow: 'cleanup', action: 'list' },
+    )
+    expect(listed.data.output).toContain(runId)
+
+    const cancelled = await WorkflowTool.call(
+      { workflow: 'cleanup', action: 'cancel', run_id: runId },
+    )
+    expect(cancelled.data.output).toContain('Workflow cancelled')
+  })
+})
--- a/packages/builtin-tools/src/tools/shared/tests/spawnMultiAgent.test.ts
+++ b/packages/builtin-tools/src/tools/shared/tests/spawnMultiAgent.test.ts
@@ -0,0 +1,54 @@
+import { afterEach, beforeEach, describe, expect, test } from 'bun:test'
+import { rmSync } from 'node:fs'
+import { tmpdir } from 'node:os'
+import { join } from 'node:path'
+import { spawnTeammate } from '../spawnMultiAgent'
+
+let tempHome: string
+let previousConfigDir: string | undefined
+
+beforeEach(() => {
+  previousConfigDir = process.env.CLAUDE_CONFIG_DIR
+  tempHome = join(tmpdir(), `spawn-multi-agent-${Date.now()}-${Math.random().toString(16).slice(2)}`)
+  process.env.CLAUDE_CONFIG_DIR = tempHome
+})
+
+afterEach(() => {
+  if (previousConfigDir === undefined) {
+    delete process.env.CLAUDE_CONFIG_DIR
+  } else {
+    process.env.CLAUDE_CONFIG_DIR = previousConfigDir
+  }
+  rmSync(tempHome, { recursive: true, force: true })
+})
+
+describe('spawnTeammate', () => {
+  test('fails before spawn side effects when the team file is missing', async () => {
+    let setAppStateCalled = false
+    const context = {
+      getAppState: () => ({
+        teamContext: undefined,
+      }),
+      setAppState: () => {
+        setAppStateCalled = true
+      },
+      options: {
+        agentDefinitions: {
+          activeAgents: [],
+        },
+      },
+    }
+
+    await expect(
+      spawnTeammate(
+        {
+          name: 'worker',
+          prompt: 'do work',
+          team_name: 'missing-team',
+        },
+        context as any,
+      ),
+    ).rejects.toThrow('Team "missing-team" does not exist')
+    expect(setAppStateCalled).toBe(false)
+  })
+})
--- a/packages/builtin-tools/src/tools/shared/spawnMultiAgent.ts
+++ b/packages/builtin-tools/src/tools/shared/spawnMultiAgent.ts
--- a/packages/modifiers-napi/src/tests/index.test.ts
+++ b/packages/modifiers-napi/src/tests/index.test.ts
@@ -0,0 +1,112 @@
+import { afterEach, beforeEach, describe, expect, mock, test } from 'bun:test'
+
+let ffiShouldThrow = false
+let nativeFlags = 0
+let dlopenCalls = 0
+
+mock.module('bun:ffi', () => ({
+  FFIType: {
+    i32: 0,
+    u64: 0,
+  },
+  dlopen: () => {
+    dlopenCalls++
+    if (ffiShouldThrow) {
+      throw new Error('ffi load failed')
+    }
+    return {
+      symbols: {
+        CGEventSourceFlagsState: () => nativeFlags,
+      },
+    }
+  },
+}))
+
+const originalPlatform = process.platform
+
+async function loadModule() {
+  return import(`../index.ts?case=${Math.random()}`)
+}
+
+beforeEach(() => {
+  ffiShouldThrow = false
+  nativeFlags = 0
+  dlopenCalls = 0
+  Object.defineProperty(process, 'platform', {
+    value: originalPlatform,
+    configurable: true,
+  })
+})
+
+afterEach(() => {
+  Object.defineProperty(process, 'platform', {
+    value: originalPlatform,
+    configurable: true,
+  })
+})
+
+describe('modifiers-napi', () => {
+  test('returns false for non-darwin platforms', async () => {
+    Object.defineProperty(process, 'platform', {
+      value: 'win32',
+      configurable: true,
+    })
+    const mod = await loadModule()
+
+    await mod.prewarm()
+    expect(dlopenCalls).toBe(0)
+    expect(mod.isModifierPressed('shift')).toBe(false)
+    expect(mod.isModifierPressed('command')).toBe(false)
+  })
+
+  test('prewarm is idempotent on darwin', async () => {
+    Object.defineProperty(process, 'platform', {
+      value: 'darwin',
+      configurable: true,
+    })
+    const mod = await loadModule()
+
+    await mod.prewarm()
+    await mod.prewarm()
+
+    expect(dlopenCalls).toBe(1)
+  })
+
+  test('returns false when ffi loading fails on darwin', async () => {
+    Object.defineProperty(process, 'platform', {
+      value: 'darwin',
+      configurable: true,
+    })
+    ffiShouldThrow = true
+    const mod = await loadModule()
+
+    await mod.prewarm()
+    expect(mod.isModifierPressed('shift')).toBe(false)
+  })
+
+  test('returns false for unknown modifier names on darwin', async () => {
+    Object.defineProperty(process, 'platform', {
+      value: 'darwin',
+      configurable: true,
+    })
+    nativeFlags = 0x20000
+    const mod = await loadModule()
+
+    await mod.prewarm()
+    expect(mod.isModifierPressed('unknown')).toBe(false)
+  })
+
+  test('uses native flag bits for known modifiers on darwin', async () => {
+    Object.defineProperty(process, 'platform', {
+      value: 'darwin',
+      configurable: true,
+    })
+    nativeFlags = 0x20000 | 0x40000
+    const mod = await loadModule()
+
+    await mod.prewarm()
+    expect(mod.isModifierPressed('shift')).toBe(true)
+    expect(mod.isModifierPressed('control')).toBe(true)
+    expect(mod.isModifierPressed('option')).toBe(false)
+  })
+})
--- a/packages/modifiers-napi/src/index.ts
+++ b/packages/modifiers-napi/src/index.ts
@@ -14,14 +14,16 @@ const modifierFlags: Record<string, number> = {
 const kCGEventSourceStateCombinedSessionState = 0;

 let cgEventSourceFlagsState: ((stateID: number) => number) | null = null;
+let ffiLoadAttempted = false;

-function loadFFI(): void {
-  if (cgEventSourceFlagsState !== null || process.platform !== "darwin") {
+async function loadFFI(): Promise<void> {
+  if (ffiLoadAttempted || process.platform !== "darwin") {
    return;
  }
+  ffiLoadAttempted = true;

  try {
-    const ffi = require("bun:ffi") as typeof import("bun:ffi");
+    const ffi = await import("bun:ffi");
    const lib = ffi.dlopen(
      `/System/Library/Frameworks/Carbon.framework/Carbon`,
      {
@@ -35,13 +37,12 @@ function loadFFI(): void {
      return Number(lib.symbols.CGEventSourceFlagsState(stateID));
    };
  } catch {
-    // If loading fails, keep the function null so isModifierPressed returns false
    cgEventSourceFlagsState = null;
  }
 }

-export function prewarm(): void {
-  loadFFI();
+export async function prewarm(): Promise<void> {
+  await loadFFI();
 }

 export function isModifierPressed(modifier: string): boolean {
@@ -49,8 +50,6 @@ export function isModifierPressed(modifier: string): boolean {
    return false;
  }

-  loadFFI();
-
  if (cgEventSourceFlagsState === null) {
    return false;
  }
--- a/packages/url-handler-napi/src/tests/index.test.ts
+++ b/packages/url-handler-napi/src/tests/index.test.ts
@@ -0,0 +1,50 @@
+import { afterEach, describe, expect, test } from 'bun:test'
+import { waitForUrlEvent } from '../index'
+
+const originalEnv = {
+  CLAUDE_CODE_URL_EVENT: process.env.CLAUDE_CODE_URL_EVENT,
+  CLAUDE_CODE_DEEP_LINK_URL: process.env.CLAUDE_CODE_DEEP_LINK_URL,
+  CLAUDE_CODE_URL: process.env.CLAUDE_CODE_URL,
+}
+const originalArgv = process.argv.slice()
+
+afterEach(() => {
+  for (const [key, value] of Object.entries(originalEnv)) {
+    if (value === undefined) {
+      delete process.env[key]
+    } else {
+      process.env[key] = value
+    }
+  }
+  process.argv = originalArgv.slice()
+})
+
+describe('waitForUrlEvent', () => {
+  test('resolves to null without a timeout', async () => {
+    await expect(waitForUrlEvent()).resolves.toBeNull()
+  })
+
+  test('resolves to null with an explicit timeout', async () => {
+    await expect(waitForUrlEvent(1)).resolves.toBeNull()
+  })
+
+  test('returns a Claude URL from environment variables', async () => {
+    process.env.CLAUDE_CODE_URL_EVENT = 'claude-cli://prompt?q=hello'
+
+    await expect(waitForUrlEvent()).resolves.toBe(
+      'claude-cli://prompt?q=hello',
+    )
+  })
+
+  test('returns a Claude URL from argv', async () => {
+    process.argv = [...originalArgv, 'claude://prompt?q=hello']
+
+    await expect(waitForUrlEvent()).resolves.toBe('claude://prompt?q=hello')
+  })
+
+  test('rejects URLs exceeding the maximum length', async () => {
+    process.env.CLAUDE_CODE_URL_EVENT = `claude-cli://${'x'.repeat(2048)}`
+
+    await expect(waitForUrlEvent()).resolves.toBeNull()
+  })
+})
--- a/packages/url-handler-napi/src/index.ts
+++ b/packages/url-handler-napi/src/index.ts
@@ -1,3 +1,48 @@
+const MAX_URL_LENGTH = 2048
+
+/**
+ * Check for a pending URL event from environment variables or CLI arguments.
+ *
+ * This is a synchronous snapshot check, not an event listener. The optional
+ * timeout parameter is retained for API compatibility but has no practical
+ * effect since process.env and process.argv do not change at runtime.
+ * Callers that need to wait for an OS-level deep link activation should use
+ * an IPC channel or platform-specific event listener instead.
+ */
 export async function waitForUrlEvent(timeoutMs?: number): Promise<string | null> {
-  return null
+  return findUrlEvent()
+}
+
+/**
+ * Checks three env var sources (set by the OS URL scheme handler or installer)
+ * and then CLI arguments for a claude:// deep link URL.
+ *
+ * Priority order:
+ * 1. CLAUDE_CODE_URL_EVENT — set by the OS URL scheme handler on activation
+ * 2. CLAUDE_CODE_DEEP_LINK_URL — set by the desktop app launcher
+ * 3. CLAUDE_CODE_URL — legacy / manual override
+ * 4. CLI arguments — e.g. `claude claude://...`
+ */
+function findUrlEvent(): string | null {
+  for (const key of [
+    'CLAUDE_CODE_URL_EVENT',
+    'CLAUDE_CODE_DEEP_LINK_URL',
+    'CLAUDE_CODE_URL',
+  ]) {
+    const value = process.env[key]
+    if (isClaudeUrl(value)) {
+      return value
+    }
+  }
+
+  const arg = process.argv.find(isClaudeUrl)
+  return arg ?? null
+}
+
+function isClaudeUrl(value: unknown): value is string {
+  return (
+    typeof value === 'string' &&
+    value.length <= MAX_URL_LENGTH &&
+    (value.startsWith('claude-cli://') || value.startsWith('claude://'))
+  )
 }