kimaki 0.30.1 → 1.0.0

package/src/message-formatting.test.ts

@@ -1,1401 +0,0 @@
-import fs from 'node:fs'
-import path from 'node:path'
-import { afterEach, describe, test, expect } from 'vitest'
-import { asDiscordQuote, asSubtext, batchChunksForDiscord, collectSessionChunks, formatBashToolTitle, formatPart, formatTaskToolTitle, formatTodoList, getTextAttachments, planAssistantTurnFlush, quotedTextFitsOneDiscordMessage, serializeEmbeds, serializePoll, serializeMessageSnapshots, sessionPartContent, shouldLeadWithBlankLine, TEXT_ATTACHMENT_INLINE_LIMIT_BYTES } from './message-formatting.js'
-import { getDataDir } from './config.js'
-import type { Collection, Embed, Message, MessageSnapshot, Poll } from 'discord.js'
-import type { Part } from '@opencode-ai/sdk/v2'
-
-describe('formatPart', () => {
-  test('callout text is returned without a diamond prefix', () => {
-    const part: Part = {
-      id: 'test',
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      text: `<callout accent="#ef4444">\n## Top priority\n- **Stripe dispute** deadline\n</callout>`,
-    }
-    expect(formatPart(part)).toMatchInlineSnapshot(`
-      "<callout accent="#ef4444">
-      ## Top priority
-      - **Stripe dispute** deadline
-      </callout>"
-    `)
-  })
-
-  test('regular text has no diamond prefix', () => {
-    const part: Part = {
-      id: 'test',
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      text: 'hello world',
-    }
-    expect(formatPart(part)).toMatchInlineSnapshot(`"hello world"`)
-  })
-
-  test('heading text has no diamond prefix', () => {
-    const part: Part = {
-      id: 'test',
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      text: '## Summary\nDone.',
-    }
-    expect(formatPart(part)).toMatchInlineSnapshot(`
-      "## Summary
-      Done."
-    `)
-  })
-
-  test('synthetic user context stays hidden', () => {
-    const part: Part = {
-      id: 'prt_branch',
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID: 'msg_user',
-      text: '\n[current git branch is main]\n',
-      synthetic: true,
-    }
-    expect(formatPart(part)).toBe('')
-  })
-
-  test('ignored plugin notices stay visible', () => {
-    const part: Part = {
-      id: 'prt_notice',
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID: 'msg_notice',
-      text: 'Subrouter: Using openai/gpt-5.6-sol because xai/grok-4.6 is rate limited.',
-      ignored: true,
-    }
-    expect(formatPart(part)).toBe(
-      'Subrouter: Using openai/gpt-5.6-sol because xai/grok-4.6 is rate limited.',
-    )
-  })
-
-})
-
-describe('asSubtext', () => {
-  test('prefixes every non-empty line', () => {
-    expect(asSubtext('hello\n\nworld')).toMatchInlineSnapshot(`
-      "-# hello
-
-      -# world"
-    `)
-  })
-
-  test('does not double-prefix', () => {
-    expect(asSubtext('-# already marked')).toMatchInlineSnapshot(`"-# already marked"`)
-  })
-})
-
-describe('asDiscordQuote', () => {
-  test('prefixes every line', () => {
-    expect(asDiscordQuote('hello\nworld')).toMatchInlineSnapshot(`
-      "> hello
-      > world"
-    `)
-  })
-
-  test('does not double-quote already quoted text', () => {
-    expect(asDiscordQuote('> already quoted')).toMatchInlineSnapshot(`"> already quoted"`)
-  })
-
-  test('prefixes blank lines and later lines, not only the first', () => {
-    expect(asDiscordQuote('I will open that file\n\nthen reply')).toMatchInlineSnapshot(`
-      "> I will open that file
-      > 
-      > then reply"
-    `)
-  })
-})
-
-describe('quotedTextFitsOneDiscordMessage', () => {
-  test('accepts a short line', () => {
-    expect(quotedTextFitsOneDiscordMessage('done')).toBe(true)
-  })
-
-  test('rejects text whose quote would split across Discord messages', () => {
-    expect(quotedTextFitsOneDiscordMessage('x'.repeat(1999))).toBe(false)
-  })
-})
-
-describe('planAssistantTurnFlush', () => {
-  function text(id: string, body: string, ended = true) {
-    return { id, type: 'text', text: body, time: ended ? { end: 1 } : undefined }
-  }
-  function tool(id: string, name?: string) {
-    return { id, type: 'tool', tool: name }
-  }
-
-  function plan(parts: Array<{ id: string; type: string; text?: string }>, mode: 'progress' | 'interactive' | 'final', throughPartId?: string) {
-    const result = planAssistantTurnFlush({ parts, mode, throughPartId })
-    return { send: result.send, hold: result.hold }
-  }
-
-  test('progress quotes completed short text immediately', () => {
-    expect(
-      plan([text('t1', 'status'), tool('tool1'), text('t2', 'answer')], 'progress'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-          {
-            "id": "t2",
-            "quoteText": true,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress quotes a lone completed short text', () => {
-    expect(plan([text('t1', 'answer')], 'progress')).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress quotes completed short text before following tools', () => {
-    expect(
-      plan([text('t1', 'status'), tool('tool1')], 'progress'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress quotes each short text that precedes a tool', () => {
-    expect(
-      plan([text('t1', 'status'), text('t2', 'next'), tool('tool1')], 'progress'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "t2",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress sends tools immediately while unfinished text is held', () => {
-    expect(
-      plan([text('t1', 'status', false), tool('tool1')], 'progress'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [
-          {
-            "id": "t1",
-            "quoteText": false,
-          },
-        ],
-        "send": [
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress holds unfinished text only when nothing else is ready', () => {
-    expect(plan([text('t1', 'status', false)], 'progress')).toMatchInlineSnapshot(`
-      {
-        "hold": [
-          {
-            "id": "t1",
-            "quoteText": false,
-          },
-        ],
-        "send": [],
-      }
-    `)
-  })
-
-  test('progress sends completed text and later tools while later text is still open', () => {
-    expect(
-      plan(
-        [text('t1', 'first'), text('t2', 'second', false), tool('tool1')],
-        'progress',
-      ),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [
-          {
-            "id": "t2",
-            "quoteText": false,
-          },
-        ],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress holds unfinished last text until it ends', () => {
-    expect(
-      plan([text('t1', 'first'), text('t2', 'second', false)], 'progress'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [
-          {
-            "id": "t2",
-            "quoteText": false,
-          },
-        ],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('final quotes earlier text and leaves the last text plain', () => {
-    expect(
-      plan([text('t1', 'status'), tool('tool1'), text('t2', 'done')], 'final'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-          {
-            "id": "t2",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('final quotes short text before a tool', () => {
-    expect(
-      plan([text('t1', 'status'), tool('tool1')], 'final'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": true,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('interactive without throughPartId sends the last text unquoted', () => {
-    expect(
-      plan([text('t1', 'ask'), tool('question1', 'question')], 'interactive'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": false,
-          },
-          {
-            "id": "question1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('interactive leaves preceding text unquoted and flushes through the interactive tool', () => {
-    expect(
-      plan([text('t1', 'ask'), tool('question1', 'question'), text('t2', 'later')], 'interactive', 'question1'),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [
-          {
-            "id": "t2",
-            "quoteText": false,
-          },
-        ],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": false,
-          },
-          {
-            "id": "question1",
-            "quoteText": false,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress does not quote earlier text longer than two lines', () => {
-    expect(
-      plan(
-        [
-          text('t1', 'line one\nline two\nline three'),
-          tool('tool1'),
-          text('t2', 'answer'),
-        ],
-        'progress',
-      ),
-    ).toMatchInlineSnapshot(`
-      {
-        "hold": [],
-        "send": [
-          {
-            "id": "t1",
-            "quoteText": false,
-          },
-          {
-            "id": "tool1",
-            "quoteText": false,
-          },
-          {
-            "id": "t2",
-            "quoteText": true,
-          },
-        ],
-      }
-    `)
-  })
-
-  test('progress still quotes two-line earlier text', () => {
-    expect(
-      plan(
-        [text('t1', 'line one\nline two'), tool('tool1'), text('t2', 'answer')],
-        'progress',
-      ).send.find((entry) => entry.id === 't1'),
-    ).toEqual({ id: 't1', quoteText: true })
-  })
-
-  test('progress keeps sleep-tool quote context after that tool is already sent', () => {
-    const parts = [text('t1', 'wait'), tool('sleep1', 'kimaki_sleep')]
-    expect(plan(parts, 'progress').send.find((entry) => entry.id === 't1')).toEqual({
-      id: 't1',
-      quoteText: false,
-    })
-    expect(plan([text('t1', 'wait')], 'progress').send.find((entry) => entry.id === 't1')).toEqual({
-      id: 't1',
-      quoteText: true,
-    })
-  })
-
-  test('progress does not quote earlier text before an interactive tool', () => {
-    expect(
-      plan(
-        [text('t1', 'ask'), tool('question1', 'question'), text('t2', 'later')],
-        'progress',
-      ).send.find((entry) => entry.id === 't1'),
-    ).toEqual({ id: 't1', quoteText: false })
-    expect(
-      plan(
-        [text('t1', 'wait'), tool('sleep1', 'kimaki_sleep'), text('t2', 'later')],
-        'progress',
-      ).send.find((entry) => entry.id === 't1'),
-    ).toEqual({ id: 't1', quoteText: false })
-    expect(
-      plan(
-        [text('t1', 'choose'), tool('buttons1', 'kimaki_action_buttons'), text('t2', 'later')],
-        'progress',
-      ).send.find((entry) => entry.id === 't1'),
-    ).toEqual({ id: 't1', quoteText: false })
-  })
-})
-
-describe('shouldLeadWithBlankLine', () => {
-  test('never leads on the first part', () => {
-    expect(
-      shouldLeadWithBlankLine({ previousKind: undefined, nextKind: 'text' }),
-    ).toBe(false)
-    expect(
-      shouldLeadWithBlankLine({ previousKind: undefined, nextKind: 'tool' }),
-    ).toBe(false)
-  })
-
-  test('leads on both text and tool transitions', () => {
-    expect(
-      shouldLeadWithBlankLine({ previousKind: 'text', nextKind: 'tool' }),
-    ).toBe(true)
-    expect(
-      shouldLeadWithBlankLine({ previousKind: 'tool', nextKind: 'text' }),
-    ).toBe(true)
-    expect(
-      shouldLeadWithBlankLine({ previousKind: 'text', nextKind: 'text' }),
-    ).toBe(false)
-    expect(
-      shouldLeadWithBlankLine({ previousKind: 'tool', nextKind: 'tool' }),
-    ).toBe(false)
-  })
-})
-
-describe('sessionPartContent', () => {
-  test('leaves the first part unchanged', () => {
-    expect(
-      sessionPartContent({ content: 'hello', leadWithBlankLine: false }),
-    ).toBe('hello')
-  })
-
-  test('prepends a newline on text and tool transitions', () => {
-    expect(
-      sessionPartContent({ content: '┣ bash ls', leadWithBlankLine: true }),
-    ).toMatchInlineSnapshot(`
-      "
-      ┣ bash ls"
-    `)
-    expect(
-      sessionPartContent({ content: 'done', leadWithBlankLine: true }),
-    ).toMatchInlineSnapshot(`
-      "
-      done"
-    `)
-  })
-})
-
-describe('collectSessionChunks', () => {
-  function textPart({
-    id,
-    text,
-    messageID,
-    synthetic,
-  }: {
-    id: string
-    text: string
-    messageID: string
-    synthetic?: boolean
-  }): Part {
-    return {
-      id,
-      type: 'text',
-      sessionID: 'ses_test',
-      messageID,
-      text,
-      synthetic,
-    }
-  }
-
-  test('skips synthetic text parts', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user' },
-          parts: [
-            textPart({
-              id: 'branch',
-              text: '\n[current git branch is main]\n',
-              messageID: 'msg_1',
-              synthetic: true,
-            }),
-            textPart({ id: 't1', text: 'working', messageID: 'msg_1' }),
-          ],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toEqual(['> working'])
-  })
-
-  test('quotes short text throughout a turn', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user' },
-          parts: [textPart({ id: 't1', text: 'I will read it', messageID: 'msg_1' })],
-        },
-        {
-          info: { role: 'assistant', id: 'msg_2', parentID: 'msg_user' },
-          parts: [textPart({ id: 't2', text: 'done', messageID: 'msg_2' })],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toMatchInlineSnapshot(`
-      [
-        "> I will read it",
-        "> done",
-      ]
-    `)
-  })
-
-  test('quotes short text before a tool', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user' },
-          parts: [
-            textPart({ id: 't1', text: 'I will read it', messageID: 'msg_1' }),
-            {
-              id: 'tool1',
-              type: 'tool',
-              sessionID: 'ses_test',
-              messageID: 'msg_1',
-              tool: 'read',
-              callID: 'call_tool1',
-              state: { status: 'completed', input: {}, output: '', title: '', metadata: {}, time: { start: 0, end: 0 } },
-            },
-          ],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toEqual([
-      '> I will read it',
-      '-# ┣ read',
-    ])
-  })
-
-  test('does not quote earlier text longer than two lines', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user' },
-          parts: [
-            textPart({
-              id: 't1',
-              text: 'line one\nline two\nline three',
-              messageID: 'msg_1',
-            }),
-          ],
-        },
-        {
-          info: { role: 'assistant', id: 'msg_2', parentID: 'msg_user' },
-          parts: [textPart({ id: 't2', text: 'done', messageID: 'msg_2' })],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toEqual([
-      'line one\nline two\nline three',
-      '> done',
-    ])
-  })
-
-  test('does not quote text flushed before a question tool', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user' },
-          parts: [
-            textPart({ id: 't1', text: 'Pick one', messageID: 'msg_1' }),
-            {
-              id: 'q1',
-              type: 'tool',
-              sessionID: 'ses_test',
-              messageID: 'msg_1',
-              tool: 'question',
-              callID: 'call_q1',
-              state: { status: 'completed', input: {}, output: '', title: '', metadata: {}, time: { start: 0, end: 0 } },
-            },
-            textPart({ id: 't2', text: 'thanks', messageID: 'msg_1' }),
-          ],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toEqual(['Pick one', '> thanks'])
-  })
-
-  test('quotes within each user turn separately', () => {
-    const { chunks } = collectSessionChunks({
-      messages: [
-        {
-          info: { role: 'assistant', id: 'msg_1', parentID: 'msg_user_1' },
-          parts: [textPart({ id: 't1', text: 'first turn', messageID: 'msg_1' })],
-        },
-        {
-          info: { role: 'user', id: 'msg_user_2' },
-          parts: [],
-        },
-        {
-          info: { role: 'assistant', id: 'msg_2', parentID: 'msg_user_2' },
-          parts: [textPart({ id: 't2', text: 'looking', messageID: 'msg_2' })],
-        },
-        {
-          info: { role: 'assistant', id: 'msg_3', parentID: 'msg_user_2' },
-          parts: [textPart({ id: 't3', text: 'second turn done', messageID: 'msg_3' })],
-        },
-      ],
-    })
-    expect(chunks.map((chunk) => chunk.content)).toMatchInlineSnapshot(`
-      [
-        "> first turn",
-        "> looking",
-        "> second turn done",
-      ]
-    `)
-  })
-})
-
-describe('batchChunksForDiscord', () => {
-  test('does not merge text chunks with tool chunks', () => {
-    expect(
-      batchChunksForDiscord([
-        { partIds: ['t1'], content: 'hello', kind: 'text' },
-        { partIds: ['b1'], content: '┣ bash ls', kind: 'tool' },
-        { partIds: ['t2'], content: 'done', kind: 'text' },
-      ]),
-    ).toMatchInlineSnapshot(`
-      [
-        {
-          "content": "hello",
-          "kind": "text",
-          "partIds": [
-            "t1",
-          ],
-        },
-        {
-          "content": "┣ bash ls",
-          "kind": "tool",
-          "partIds": [
-            "b1",
-          ],
-        },
-        {
-          "content": "done",
-          "kind": "text",
-          "partIds": [
-            "t2",
-          ],
-        },
-      ]
-    `)
-  })
-})
-
-describe('formatTaskToolTitle', () => {
-  function taskPart({
-    status,
-    input = {},
-    title = '',
-    sessionId,
-  }: {
-    status: 'running' | 'completed'
-    input?: { description?: string; subagent_type?: string }
-    title?: string
-    sessionId?: string
-  }): Extract<Part, { type: 'tool' }> {
-    const base = {
-      id: 'prt_task',
-      type: 'tool' as const,
-      tool: 'task',
-      callID: 'call_task',
-      sessionID: 'ses_parent',
-      messageID: 'msg_assistant',
-    }
-    if (status === 'completed') {
-      return {
-        ...base,
-        state: {
-          status,
-          input,
-          output: '',
-          title,
-          metadata: sessionId ? { sessionId } : {},
-          time: { start: 1, end: 2 },
-        },
-      }
-    }
-    return {
-      ...base,
-      state: {
-        status,
-        input,
-        title,
-        metadata: sessionId ? { sessionId } : {},
-        time: { start: 1 },
-      },
-    }
-  }
-
-  test('uses the running task title when OpenAI omits the description', () => {
-    expect(
-      formatTaskToolTitle(
-        taskPart({
-          status: 'running',
-          input: { subagent_type: 'general' },
-          title: 'Classify pending changes',
-          sessionId: 'ses_child',
-        }),
-      ),
-    ).toMatchInlineSnapshot(`"-# ┣ general **Classify pending changes**"`)
-  })
-
-  test('prefers input description on running parts', () => {
-    expect(
-      formatTaskToolTitle(
-        taskPart({
-          status: 'running',
-          input: { description: 'inspect repo', subagent_type: 'explore' },
-          title: 'ignored title',
-          sessionId: 'ses_child',
-        }),
-      ),
-    ).toMatchInlineSnapshot(`"-# ┣ explore **inspect repo**"`)
-  })
-
-  test('does not format completed parts so Discord does not post the line at the end', () => {
-    expect(
-      formatTaskToolTitle(
-        taskPart({
-          status: 'completed',
-          input: { subagent_type: 'general' },
-          title: 'Classify pending changes',
-          sessionId: 'ses_child',
-        }),
-      ),
-    ).toBe('')
-  })
-
-  test('shows queued task calls before the child session exists', () => {
-    expect(
-      formatTaskToolTitle(
-        taskPart({
-          status: 'running',
-          input: {
-            description: 'audit customer pages',
-            subagent_type: 'general',
-          },
-        }),
-      ),
-    ).toMatchInlineSnapshot(`"-# ┣ general **audit customer pages**"`)
-  })
-})
-
-describe('formatBashToolTitle', () => {
-  test('short single-line command shown in full', () => {
-    expect(formatBashToolTitle({ command: 'echo hello' })).toMatchInlineSnapshot(`" _echo hello_"`)
-  })
-
-  test('multiline command without description truncates to first line', () => {
-    expect(
-      formatBashToolTitle({ command: 'echo hello\necho world\necho done' }),
-    ).toMatchInlineSnapshot(`" _echo hello…_"`)
-  })
-
-  test('long single-line command is truncated with ellipsis', () => {
-    const longCommand = 'a'.repeat(150)
-    const result = formatBashToolTitle({ command: longCommand })
-    expect(result).toContain('…')
-    expect(result.length).toBeLessThan(150)
-  })
-
-  test('description is preferred over truncated command when present', () => {
-    expect(
-      formatBashToolTitle({
-        command: 'echo hello\necho world',
-        description: 'Print greeting',
-      }),
-    ).toMatchInlineSnapshot(`" _Print greeting_"`)
-  })
-
-  test('50-char single-line command is shown in full', () => {
-    const command = 'a'.repeat(50)
-    expect(formatBashToolTitle({ command, description: 'Ignored' })).toBe(
-      ` _${command}_`,
-    )
-  })
-
-  test('51-char single-line command uses description', () => {
-    expect(
-      formatBashToolTitle({
-        command: 'a'.repeat(51),
-        description: 'Run long command',
-      }),
-    ).toMatchInlineSnapshot(`" _Run long command_"`)
-  })
-
-  test('51-char single-line command uses summary when description is missing', () => {
-    expect(
-      formatBashToolTitle({
-        command: 'a'.repeat(51),
-        summary: 'Install packages',
-      }),
-    ).toMatchInlineSnapshot(`" _Install packages_"`)
-  })
-
-  test('description wins over summary for long commands', () => {
-    expect(
-      formatBashToolTitle({
-        command: 'a'.repeat(51),
-        description: 'Preferred label',
-        summary: 'Ignored label',
-      }),
-    ).toMatchInlineSnapshot(`" _Preferred label_"`)
-  })
-
-  test('stateTitle used as last resort', () => {
-    expect(
-      formatBashToolTitle({ command: '', stateTitle: 'Running tests' }),
-    ).toMatchInlineSnapshot(`" _Running tests_"`)
-  })
-
-  test('empty inputs return empty string', () => {
-    expect(formatBashToolTitle({ command: '' })).toBe('')
-  })
-
-  test('leading blank line skipped, uses first meaningful line', () => {
-    expect(
-      formatBashToolTitle({ command: '\npnpm test\npnpm build' }),
-    ).toMatchInlineSnapshot(`" _pnpm test…_"`)
-  })
-
-  test('whitespace-only first line skipped', () => {
-    expect(
-      formatBashToolTitle({ command: '   \npnpm test' }),
-    ).toMatchInlineSnapshot(`" _pnpm test…_"`)
-  })
-
-  test('no description field (new opencode) with multiline command', () => {
-    // This is the exact scenario that was broken: opencode removed `description`
-    // from the bash tool schema, so multiline commands rendered as just "┣ bash"
-    const command = 'git diff HEAD~1 --stat && git log --oneline -5'
-    expect(formatBashToolTitle({ command: command + '\n' + 'echo done' })).toMatchInlineSnapshot(
-      `" _git diff HEAD\\~1 --stat && git log --oneline -5…_"`,
-    )
-  })
-})
-
-describe('formatTodoList', () => {
-  test('formats active todo with number, dot, and two spaces', () => {
-    const part: Part = {
-      id: 'test',
-      type: 'tool',
-      tool: 'todowrite',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      callID: 'call_test',
-      state: {
-        status: 'completed',
-        input: {
-          todos: [
-            { content: 'First task', status: 'completed' },
-            { content: 'Second task', status: 'in_progress' },
-            { content: 'Third task', status: 'pending' },
-          ],
-        },
-        output: '',
-        title: 'todowrite',
-        metadata: {},
-        time: { start: 0, end: 0 },
-      },
-    }
-
-    expect(formatTodoList(part)).toMatchInlineSnapshot(`"2.  **second task**"`)
-  })
-
-  test('formats double digit todo numbers', () => {
-    const todos = Array.from({ length: 12 }, (_, i) => ({
-      content: `Task ${i + 1}`,
-      status: i === 11 ? 'in_progress' : 'completed',
-    }))
-
-    const part: Part = {
-      id: 'test',
-      type: 'tool',
-      tool: 'todowrite',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      callID: 'call_test',
-      state: {
-        status: 'completed',
-        input: { todos },
-        output: '',
-        title: 'todowrite',
-        metadata: {},
-        time: { start: 0, end: 0 },
-      },
-    }
-
-    expect(formatTodoList(part)).toMatchInlineSnapshot(`"12.  **task 12**"`)
-  })
-
-  test('lowercases first letter of content', () => {
-    const part: Part = {
-      id: 'test',
-      type: 'tool',
-      tool: 'todowrite',
-      sessionID: 'ses_test',
-      messageID: 'msg_test',
-      callID: 'call_test',
-      state: {
-        status: 'completed',
-        input: {
-          todos: [{ content: 'Fix the bug', status: 'in_progress' }],
-        },
-        output: '',
-        title: 'todowrite',
-        metadata: {},
-        time: { start: 0, end: 0 },
-      },
-    }
-
-    expect(formatTodoList(part)).toMatchInlineSnapshot(`"1.  **fix the bug**"`)
-  })
-})
-
-describe('serializeEmbeds', () => {
-  function fakeEmbed(data: {
-    title?: string
-    description?: string
-    url?: string
-    author?: { name: string }
-    footer?: { text: string }
-    fields?: Array<{ name: string; value: string; inline?: boolean }>
-  }): Embed {
-    return {
-      title: data.title ?? null,
-      description: data.description ?? null,
-      url: data.url ?? null,
-      author: data.author ?? null,
-      footer: data.footer ?? null,
-      fields: data.fields ?? [],
-    } as unknown as Embed
-  }
-
-  test('serializes a full embed with all fields', () => {
-    const embeds = [
-      fakeEmbed({
-        author: { name: 'GitHub' },
-        title: 'PR #42: Fix auth timeout',
-        url: 'https://github.com/org/repo/pull/42',
-        description: 'Fixes the retry logic so tokens refresh before expiry.',
-        fields: [
-          { name: 'Status', value: 'Open' },
-          { name: 'Reviewers', value: 'alice, bob' },
-        ],
-        footer: { text: 'Last updated 2h ago' },
-      }),
-    ]
-    expect(serializeEmbeds(embeds)).toMatchInlineSnapshot(`
-      "<embed>
-      Author: GitHub
-      Title: PR #42: Fix auth timeout
-      URL: https://github.com/org/repo/pull/42
-      Fixes the retry logic so tokens refresh before expiry.
-      Status: Open
-      Reviewers: alice, bob
-      Footer: Last updated 2h ago
-      </embed>"
-    `)
-  })
-
-  test('serializes description-only embed (link preview)', () => {
-    const embeds = [
-      fakeEmbed({
-        title: 'Example Site',
-        url: 'https://example.com',
-        description: 'An example website for testing.',
-      }),
-    ]
-    expect(serializeEmbeds(embeds)).toMatchInlineSnapshot(`
-      "<embed>
-      Title: Example Site
-      URL: https://example.com
-      An example website for testing.
-      </embed>"
-    `)
-  })
-
-  test('returns empty string for no embeds', () => {
-    expect(serializeEmbeds([])).toBe('')
-  })
-
-  test('skips embeds with no text content', () => {
-    // An embed with only an image and no text fields
-    const embeds = [fakeEmbed({})]
-    expect(serializeEmbeds(embeds)).toBe('')
-  })
-
-  test('serializes multiple embeds', () => {
-    const embeds = [
-      fakeEmbed({ title: 'First', description: 'one' }),
-      fakeEmbed({ title: 'Second', description: 'two' }),
-    ]
-    expect(serializeEmbeds(embeds)).toMatchInlineSnapshot(`
-      "<embed>
-      Title: First
-      one
-      </embed>
-
-      <embed>
-      Title: Second
-      two
-      </embed>"
-    `)
-  })
-})
-
-// Helper to create a fake Map-like Collection for tests
-function fakeCollection<K, V>(entries: [K, V][]): Collection<K, V> {
-  const map = new Map(entries)
-  return {
-    size: map.size,
-    [Symbol.iterator]: map[Symbol.iterator].bind(map),
-  } as unknown as Collection<K, V>
-}
-
-describe('serializePoll', () => {
-  function fakePoll(data: {
-    question: string
-    answers: Array<{ id: number; text: string | null }>
-  }): Poll {
-    return {
-      question: { text: data.question },
-      answers: fakeCollection(
-        data.answers.map((a) => [a.id, { text: a.text }]),
-      ),
-    } as unknown as Poll
-  }
-
-  test('serializes a poll with question and answers', () => {
-    const poll = fakePoll({
-      question: 'Which framework?',
-      answers: [
-        { id: 1, text: 'React' },
-        { id: 2, text: 'Vue' },
-        { id: 3, text: 'Svelte' },
-      ],
-    })
-    expect(serializePoll(poll)).toMatchInlineSnapshot(`
-      "<poll>
-      Question: Which framework?
-      - React
-      - Vue
-      - Svelte
-      </poll>"
-    `)
-  })
-
-  test('returns empty string for null poll', () => {
-    expect(serializePoll(null)).toBe('')
-  })
-
-  test('skips answers with no text', () => {
-    const poll = fakePoll({
-      question: 'Pick one',
-      answers: [
-        { id: 1, text: 'Option A' },
-        { id: 2, text: null },
-      ],
-    })
-    expect(serializePoll(poll)).toMatchInlineSnapshot(`
-      "<poll>
-      Question: Pick one
-      - Option A
-      </poll>"
-    `)
-  })
-})
-
-describe('serializeMessageSnapshots', () => {
-  function fakeSnapshot(data: {
-    content?: string
-    embeds?: Embed[]
-  }): MessageSnapshot {
-    return {
-      content: data.content ?? '',
-      embeds: data.embeds ?? [],
-    } as unknown as MessageSnapshot
-  }
-
-  function fakeEmbed(data: {
-    title?: string
-    description?: string
-    url?: string
-    author?: { name: string }
-    footer?: { text: string }
-    fields?: Array<{ name: string; value: string }>
-  }): Embed {
-    return {
-      title: data.title ?? null,
-      description: data.description ?? null,
-      url: data.url ?? null,
-      author: data.author ?? null,
-      footer: data.footer ?? null,
-      fields: data.fields ?? [],
-    } as unknown as Embed
-  }
-
-  test('serializes a forwarded message with content', () => {
-    const snapshots = fakeCollection<string, MessageSnapshot>([
-      ['1', fakeSnapshot({ content: 'Hello from another channel' })],
-    ])
-    expect(serializeMessageSnapshots(snapshots)).toMatchInlineSnapshot(`
-      "<forwarded-message>
-      Hello from another channel
-      </forwarded-message>"
-    `)
-  })
-
-  test('serializes forwarded message with content and embeds', () => {
-    const snapshots = fakeCollection<string, MessageSnapshot>([
-      [
-        '1',
-        fakeSnapshot({
-          content: 'Check this out',
-          embeds: [fakeEmbed({ title: 'Link Preview', description: 'A cool site' })],
-        }),
-      ],
-    ])
-    expect(serializeMessageSnapshots(snapshots)).toMatchInlineSnapshot(`
-      "<forwarded-message>
-      Check this out
-
-      <embed>
-      Title: Link Preview
-      A cool site
-      </embed>
-      </forwarded-message>"
-    `)
-  })
-
-  test('returns empty string for no snapshots', () => {
-    const empty = fakeCollection<string, MessageSnapshot>([])
-    expect(serializeMessageSnapshots(empty)).toBe('')
-  })
-
-  test('skips snapshots with no content', () => {
-    const snapshots = fakeCollection<string, MessageSnapshot>([
-      ['1', fakeSnapshot({})],
-    ])
-    expect(serializeMessageSnapshots(snapshots)).toBe('')
-  })
-
-  test('serializes multiple forwarded messages', () => {
-    const snapshots = fakeCollection<string, MessageSnapshot>([
-      ['1', fakeSnapshot({ content: 'First forwarded' })],
-      ['2', fakeSnapshot({ content: 'Second forwarded' })],
-    ])
-    expect(serializeMessageSnapshots(snapshots)).toMatchInlineSnapshot(`
-      "<forwarded-message>
-      First forwarded
-      </forwarded-message>
-
-      <forwarded-message>
-      Second forwarded
-      </forwarded-message>"
-    `)
-  })
-})
-
-describe('getTextAttachments', () => {
-  const originalFetch = globalThis.fetch
-
-  afterEach(() => {
-    globalThis.fetch = originalFetch
-  })
-
-  function stubFetch(impl: (url: string) => Promise<Response>) {
-    const fetchFn = async (input: string | URL | Request) => impl(String(input))
-    fetchFn.preconnect = () => {}
-    globalThis.fetch = fetchFn
-  }
-
-  function messageWithAttachments(
-    attachments: Array<{
-      id: string
-      name: string
-      contentType: string
-      url: string
-      size: number
-    }>,
-  ) {
-    return {
-      attachments: new Map(attachments.map((attachment) => {
-        return [attachment.id, attachment]
-      })),
-    } as unknown as Message
-  }
-
-  function snapshotAttachments(result: string) {
-    return result.replaceAll(getDataDir(), '<dataDir>')
-  }
-
-  test('inlines small text files and saves them locally', async () => {
-    stubFetch(async () => {
-      return new Response('hello from file', { status: 200 })
-    })
-
-    const result = await getTextAttachments(
-      messageWithAttachments([
-        {
-          id: 'att1',
-          name: 'notes.txt',
-          contentType: 'text/plain',
-          url: 'https://cdn.example/notes.txt',
-          size: 16,
-        },
-      ]),
-    )
-
-    const savedPath = path.join(getDataDir(), 'attachments', 'att1-notes.txt')
-    expect(fs.readFileSync(savedPath, 'utf8')).toBe('hello from file')
-    expect(result).toContain('https://cdn.example/notes.txt')
-    expect(result).toContain(savedPath)
-    expect(snapshotAttachments(result)).toMatchInlineSnapshot(`
-      "<attachment filename="notes.txt" mime="text/plain" size="16" url="https://cdn.example/notes.txt" path="<dataDir>/attachments/att1-notes.txt">
-      hello from file
-      </attachment>"
-    `)
-  })
-
-  test('saves large text files locally without inlining contents', async () => {
-    stubFetch(async () => {
-      return new Response('SHOULD NOT BE INLINED', { status: 200 })
-    })
-
-    const result = await getTextAttachments(
-      messageWithAttachments([
-        {
-          id: 'att2',
-          name: 'huge.log',
-          contentType: 'text/plain',
-          url: 'https://cdn.discordapp.com/attachments/1/2/huge.log',
-          size: TEXT_ATTACHMENT_INLINE_LIMIT_BYTES + 1,
-        },
-      ]),
-    )
-
-    const savedPath = path.join(getDataDir(), 'attachments', 'att2-huge.log')
-    expect(fs.readFileSync(savedPath, 'utf8')).toBe('SHOULD NOT BE INLINED')
-    expect(result).not.toContain('SHOULD NOT BE INLINED')
-    expect(result).toContain('https://cdn.discordapp.com/attachments/1/2/huge.log')
-    expect(result).toContain(savedPath)
-    expect(result).toContain('text/plain')
-    expect(result).toContain(String(TEXT_ATTACHMENT_INLINE_LIMIT_BYTES + 1))
-    expect(snapshotAttachments(result)).toMatchInlineSnapshot(`
-      "<attachment filename="huge.log" mime="text/plain" size="65537" url="https://cdn.discordapp.com/attachments/1/2/huge.log" path="<dataDir>/attachments/att2-huge.log" large="true">
-      This file is large (64 KB, text/plain). Contents were not inlined to save context. Read the local path.
-      </attachment>"
-    `)
-  })
-
-  test('still inlines large prompt.md attachments from kimaki send', async () => {
-    stubFetch(async () => {
-      return new Response('the actual long prompt', { status: 200 })
-    })
-
-    const result = await getTextAttachments(
-      messageWithAttachments([
-        {
-          id: 'att3',
-          name: 'prompt.md',
-          contentType: 'text/markdown',
-          url: 'https://cdn.example/prompt.md',
-          size: TEXT_ATTACHMENT_INLINE_LIMIT_BYTES + 1,
-        },
-      ]),
-    )
-
-    const savedPath = path.join(getDataDir(), 'attachments', 'att3-prompt.md')
-    expect(fs.readFileSync(savedPath, 'utf8')).toBe('the actual long prompt')
-    expect(result).toContain('the actual long prompt')
-    expect(result).toContain('https://cdn.example/prompt.md')
-    expect(result).toContain(savedPath)
-    expect(snapshotAttachments(result)).toMatchInlineSnapshot(`
-      "<attachment filename="prompt.md" mime="text/markdown" size="65537" url="https://cdn.example/prompt.md" path="<dataDir>/attachments/att3-prompt.md">
-      the actual long prompt
-      </attachment>"
-    `)
-  })
-
-  test('keeps small files inlined next to large file references', async () => {
-    stubFetch(async (url) => {
-      if (url.includes('small.txt')) {
-        return new Response('tiny', { status: 200 })
-      }
-      return new Response('SHOULD NOT BE INLINED', { status: 200 })
-    })
-
-    const result = await getTextAttachments(
-      messageWithAttachments([
-        {
-          id: 'att4',
-          name: 'small.txt',
-          contentType: 'text/plain',
-          url: 'https://cdn.example/small.txt',
-          size: 4,
-        },
-        {
-          id: 'att5',
-          name: 'dump.json',
-          contentType: 'application/json',
-          url: 'https://cdn.example/dump.json',
-          size: TEXT_ATTACHMENT_INLINE_LIMIT_BYTES + 50,
-        },
-      ]),
-    )
-
-    expect(result).toContain('tiny')
-    expect(result).not.toContain('SHOULD NOT BE INLINED')
-    expect(result).toContain('https://cdn.example/small.txt')
-    expect(result).toContain('https://cdn.example/dump.json')
-    expect(result).toContain(path.join(getDataDir(), 'attachments', 'att4-small.txt'))
-    expect(result).toContain(path.join(getDataDir(), 'attachments', 'att5-dump.json'))
-    expect(result).toContain('application/json')
-    expect(snapshotAttachments(result)).toMatchInlineSnapshot(`
-      "<attachment filename="small.txt" mime="text/plain" size="4" url="https://cdn.example/small.txt" path="<dataDir>/attachments/att4-small.txt">
-      tiny
-      </attachment>
-
-      <attachment filename="dump.json" mime="application/json" size="65586" url="https://cdn.example/dump.json" path="<dataDir>/attachments/att5-dump.json" large="true">
-      This file is large (64 KB, application/json). Contents were not inlined to save context. Read the local path.
-      </attachment>"
-    `)
-  })
-})