continuedev/continue · error · Error

VertexAI in express mode (apiKey only) cannot be configured

Error message

VertexAI in express mode (apiKey only) cannot be configured with region, projectId, keyFile, or keyJson

What it means

VertexAI setupAuthentication runs in express mode when apiKey is set; it forbids mixing express mode with standard-mode credentials. If any of region, projectId, keyFile, or keyJson is also present, this error is thrown from the constructor.

Source

Thrown at packages/openai-adapters/src/apis/VertexAI.ts:80

      this.genAI = new GoogleGenAI({ apiKey });
    } else if (env?.projectId && env?.region) {
      this.genAI = new GoogleGenAI({
        vertexai: true,
        project: env.projectId,
        location: env.region,
      });
    }
  }

  private setupAuthentication(): void {
    const { apiKey, env } = this.config;
    const { region, projectId, keyFile, keyJson } = env || {};

    // Validate authentication configuration
    if (apiKey) {
      // Express mode validation
      if (region || projectId || keyFile || keyJson) {
        throw new Error(
          "VertexAI in express mode (apiKey only) cannot be configured with region, projectId, keyFile, or keyJson",
        );
      }
    } else {
      // Standard mode validation
      if (!region || !projectId) {
        throw new Error(
          "region and projectId are required for VertexAI (when not using express/apiKey mode)",
        );
      }
      if (keyFile && keyJson) {
        throw new Error(
          "VertexAI credentials can be configured with either keyFile or keyJson but not both",
        );
      }
    }

    // Set up authentication client

View on GitHub (pinned to 5522c6f44c)

Solutions

1. Remove region, projectId, keyFile, and keyJson from the Vertex config and keep only apiKey (express mode)
2. Or drop apiKey and provide region + projectId (plus keyFile or keyJson) for standard service-account mode
3. Audit env vars like GOOGLE_VERTEX_REGION, GOOGLE_CLOUD_PROJECT that leak into the adapter config

Example fix

// before
new VertexAIApi({ apiKey, env: { region: 'us-central1', projectId: 'my-proj' } })
// after
new VertexAIApi({ apiKey })

const { apiKey, region, projectId, keyFile, keyJson } = cfg; if (apiKey && (region || projectId || keyFile || keyJson)) throw new Error('Remove standard-mode fields or the apiKey');

const isCleanExpressConfig = (c: VertexConfig) => !!c.apiKey && !c.env?.region && !c.env?.projectId && !c.env?.keyFile && !c.env?.keyJson;

try { new VertexAIApi(cfg); } catch (e) { if (e.message.includes('express mode')) throw new ConfigError('Pick one VertexAI auth mode'); throw e; }

When it happens

Trigger: Constructing VertexAIApi with apiKey plus region/projectId/keyFile/keyJson in env/config (e.g. GOOGLE_VERTEX_REGION set alongside VERTEX_API_KEY).

Common situations: Shared .env files containing leftover service-account variables while switching to express API-key auth; CI environments injecting GOOGLE_CLOUD_PROJECT globally; combining examples from both auth modes.

Related errors

• region and projectId are required for VertexAI (when not usi
• URL is not defined in params
• Response body is null
• Profile ${profileId} not found
• No workspace directories found. Make sure you've opened a fo

AI-assisted analysis of continuedev/continue@5522c6f44c (2026-08-27). Data as JSON: /api/errors/fa4f2962b100b706. Report an issue: GitHub.