linshenkx/prompt-optimizer · error · EvaluationModelError

${modelKey}

Error message

${modelKey}

What it means

EvaluationModelError is thrown by validateModel when modelManager.getModel(modelKey) resolves to a falsy value — i.e. the evaluationModelKey passed on the request does not correspond to any registered/loaded text model. The message is just the model key; the structured context is in error.params.context and the code is MODEL_NOT_FOUND.

Source

Thrown at packages/core/src/services/evaluation/service.ts:559

          throw new EvaluationValidationError('Workspace prompt must not be empty.');
        }
        if (!request.iterateRequirement?.trim()) {
          throw new EvaluationValidationError('Iteration requirement must not be empty.');
        }
        break;

      default:
        throw new EvaluationValidationError(`Unknown evaluation type: ${(request as any).type}`);
    }
  }

  /**
   * 验证评估模型
   */
  private async validateModel(modelKey: string): Promise<TextModelConfig> {
    const model = await this.modelManager.getModel(modelKey);
    if (!model) {
      throw new EvaluationModelError(modelKey);
    }
    return model;
  }

  /**
   * 获取评估模板
   */
  private async getEvaluationTemplate(type: EvaluationType, mode: EvaluationModeConfig): Promise<Template> {
    const templateId = this.getTemplateId(type, mode);

    try {
      const template = await this.templateManager.getTemplate(templateId);
      if (!template?.content) {
        throw new EvaluationTemplateError(templateId);
      }
      return template;
    } catch (error) {
      if (error instanceof EvaluationTemplateError) {

View on GitHub (pinned to 3e677b1d9f)

Solutions

1. Check that evaluationModelKey exactly matches a key registered in ModelManager (list available models first)
2. Register/load the intended model configuration before calling evaluate
3. Await model-manager initialization at startup if models load async
4. Update persisted user preferences that reference deleted model keys

Example fix

// before
await svc.evaluate({ evaluationModelKey: 'gpt4-turbo', ... });

// after
const models = await modelManager.listModels();
const key = models.some(m => m.key === 'gpt-4-turbo') ? 'gpt-4-turbo' : models[0].key;
await svc.evaluate({ evaluationModelKey: key, ... });

const model = await modelManager.getModel(key);
if (!model) throw new Error(`Model '${key}' not registered; available: ...`);
await svc.evaluate({ ...req, evaluationModelKey: key });

try { await svc.evaluate(req); } catch (e) { if (e instanceof EvaluationModelError) { await refreshModelList(); retryWithDefaultModel(); } else throw e; }

When it happens

Trigger: evaluate({evaluationModelKey:'gpt-4x', ...}) where no model with that key is configured in the ModelManager; model removed/renamed after the request was built; model provider not initialized at startup.

Common situations: Model config file missing or key renamed between environments; user-selected model from an outdated dropdown list; model manager loaded asynchronously and evaluate called before models registered; typo in the key.

Related errors

• ${templateId}
• Evaluation model key must not be empty.
• Function mode must not be empty.
• Sub mode must not be empty.
• Result evaluation snapshot testCaseId must match testCase.id

AI-assisted analysis of linshenkx/prompt-optimizer@3e677b1d9f (2026-08-27). Data as JSON: /api/errors/9e29ab731d51d54c. Report an issue: GitHub.