linshenkx/prompt-optimizer · error · RequestConfigError

Message content must be a string

Error message

Message content must be a string

What it means

Thrown by validateMessages when a message's content is present but is not a string (e.g. a number, object, or array). The adapter's request serialization expects plain string content; multimodal/content-part arrays are not supported by this validation path. It is a client-side RequestConfigError raised before any network request.

Source

Thrown at packages/core/src/services/llm/adapters/abstract-adapter.ts:197

    if (!Array.isArray(messages)) {
      throw new RequestConfigError('Messages must be an array')
    }

    if (messages.length === 0) {
      throw new RequestConfigError('Messages array cannot be empty')
    }

    for (const msg of messages) {
      if (!msg.role || !msg.content) {
        throw new RequestConfigError('Each message must have role and content')
      }

      if (!['system', 'user', 'assistant', 'tool'].includes(msg.role)) {
        throw new RequestConfigError(`Invalid message role: ${msg.role}`)
      }

      if (typeof msg.content !== 'string') {
        throw new RequestConfigError('Message content must be a string')
      }
    }
  }

  protected validateImageUnderstandingRequest(request: ImageUnderstandingRequest): void {
    if (!request || typeof request !== 'object') {
      throw new RequestConfigError('Image understanding request cannot be empty')
    }

    if (typeof request.userPrompt !== 'string' || !request.userPrompt.trim()) {
      throw new RequestConfigError('Image understanding user prompt cannot be empty')
    }

    if (!Array.isArray(request.images) || request.images.length === 0) {
      throw new RequestConfigError('Image understanding request requires at least one image')
    }

    request.images.forEach((image, index) => {

View on GitHub (pinned to 3e677b1d9f)

Solutions

1. Stringify content before sending: String(value) or JSON.stringify for structured data
2. For multimodal/image input, use the dedicated sendImageUnderstanding API instead of message content arrays
3. Type messages strictly so content: string is enforced at compile time

Example fix

// before
messages = [{ role: 'user', content: [{ type: 'text', text: 'hi' }] }]

// after
messages = [{ role: 'user', content: 'hi' }]

messages.forEach(m => { if (typeof m.content !== 'string') m.content = String(m.content) })

function hasStringContent(m: unknown): m is Message {
  return !!m && typeof (m as Message).content === 'string'
}

null

When it happens

Trigger: Passing content as a number (content: 42), a content-parts array (content: [{type:'text',...}]), or an object; JSON.parse results where content ended up as a non-string value.

Common situations: Reusing OpenAI multimodal content arrays; interpolating computed values into content without String(); loose typing where Message['content'] is any.

Related errors

• Messages array cannot be empty
• Each message must have role and content
• Invalid message role: ${msg.role}
• Image at index ${index} is invalid
• Model provider metadata cannot be empty

AI-assisted analysis of linshenkx/prompt-optimizer@3e677b1d9f (2026-08-27). Data as JSON: /api/errors/218f0af732bd608e. Report an issue: GitHub.