linshenkx/prompt-optimizer · error · ContextError

CONTEXT_ERROR_CODES.IMPORT_FORMAT_ERROR

CONTEXT_ERROR_CODES.IMPORT_FORMAT_ERROR

Error message

Invalid context bundle data

What it means

Last-resort failure in resolvePromptObject: every strategy to obtain a JSON object from the model response failed — direct parse, extraction of embedded JSON, etc. (earlier attempts are swallowed by the catch with 'ignore and fall through'). The model output contained no usable JSON object at all.

Solutions

1. Increase maxOutputTokens so the JSON isn't truncated
2. Reinforce JSON-only output in the system prompt and keep responseMimeType 'application/json'
3. Retry once — LLM JSON compliance is nondeterministic
4. Log the raw response to see which extraction strategy failed and adjust the slice logic (e.g. handle fenced ```json blocks)

Example fix

// before
const systemPrompt = 'Extract the prompt.'

// after
const systemPrompt =
  'Extract the prompt. Respond with ONLY a single valid JSON object. ' +
  'Do not wrap it in markdown. Do not add any explanation text.'
// and ensure maxOutputTokens is large enough for the full object

function containsJsonObject(text: string): boolean {
  const match = text.match(/\{[\s\S]*}/)
  if (!match) return false
  try {
    return isRecord(JSON.parse(match[0]))
  } catch {
    return false
  }
}

function isRecord(v: unknown): v is Record<string, unknown> {
  return typeof v === 'object' && v !== null && !Array.isArray(v)
}

try {
  const obj = await extractor.resolvePromptObject(rawResponse)
} catch (error) {
  if (error instanceof Error && error.message.includes('missing a usable JSON prompt object')) {
    return buildPromptFromPlainText(rawResponse) // degrade to plain-text prompt
  }
  throw error
}

When it happens

Trigger: Calling promptObject / resolvePromptObject when the model returns prose without any JSON, markdown code fences that don't contain a complete object, or truncated output that can't be parsed from any extracted slice.

Common situations: Model ignores JSON mode instructions; output truncated by maxOutputTokens; model answers conversationally ('Sure, here is...') without braces; long responses cut off mid-object.

Related errors

• Optimized text must be a string
• CONTEXT_ERROR_CODES.INVALID_STORE
• Text comparison calculation failed

AI-assisted analysis of linshenkx/prompt-optimizer@3e677b1d9f (2026-08-27). Data as JSON: /api/errors/03248b2c62247e12. Report an issue: GitHub.

Appendix: source

Thrown at packages/core/src/services/context/electron-proxy.ts:105

  }

  // === 导入导出 ===
  async exportAll(): Promise<ContextBundle> {
    return this.api.exportAll();
  }

  async importAll(bundle: ContextBundle, mode: ImportMode): Promise<ImportResult> {
    return this.api.importAll(safeSerializeForIPC(bundle), mode);
  }

  // === IImportExportable 实现 ===
  async exportData(): Promise<ContextBundle> {
    return this.exportAll();
  }

  async importData(data: any): Promise<void> {
    if (!(await this.validateData(data))) {
      throw new ContextError(CONTEXT_ERROR_CODES.IMPORT_FORMAT_ERROR, 'Invalid context bundle data');
    }
    await this.importAll(data as ContextBundle, 'replace');
  }

  async getDataType(): Promise<string> {
    return this.api.getDataType ? this.api.getDataType() : Promise.resolve('context-bundle');
  }

  async validateData(data: any): Promise<boolean> {
    return this.api.validateData 
      ? this.api.validateData(safeSerializeForIPC(data))
      : Promise.resolve(!!(data?.type && data?.type === 'context-bundle'));
  }
}

View on GitHub (pinned to 3e677b1d9f)