abhigyanpatwari/GitNexus · error · Error

initEmbedder() should not be called in HTTP mode. Use embedT

Error message

initEmbedder() should not be called in HTTP mode. Use embedText()/embedBatch() which handle HTTP transparently.

What it means

Thrown by initEmbedder() when isHttpMode() is true — i.e. both GITNEXUS_EMBEDDING_URL and GITNEXUS_EMBEDDING_MODEL are set. In HTTP mode the local ONNX Runtime pipeline is intentionally never loaded; embedText() and embedBatch() detect HTTP mode internally and dispatch to httpEmbed() transparently. Calling initEmbedder() would bypass that and try to load a native model, so it is rejected at the top of the function before any native import.

Source

Thrown at gitnexus/src/core/embeddings/embedder.ts:74

 */
export const getCurrentDevice = (): 'dml' | 'cuda' | 'cpu' | 'wasm' | null => currentDevice;

/**
 * Initialize the embedding model
 * Uses singleton pattern - only loads once, subsequent calls return cached instance
 *
 * @param onProgress - Optional callback for model download progress
 * @param config - Optional configuration override
 * @param forceDevice - Force a specific device
 * @returns Promise resolving to the embedder pipeline
 */
export const initEmbedder = async (
  onProgress?: ModelProgressCallback,
  config: Partial<EmbeddingConfig> = {},
  forceDevice?: 'dml' | 'cuda' | 'cpu' | 'wasm',
): Promise<FeatureExtractionPipeline> => {
  if (isHttpMode()) {
    throw new Error(
      'initEmbedder() should not be called in HTTP mode. ' +
        'Use embedText()/embedBatch() which handle HTTP transparently.',
    );
  }

  // Fail fast on platforms where the bundled native ONNX Runtime binding is not
  // shipped (macOS Intel, #1515). Must run before any transformers.js /
  // onnxruntime-node import or resolution — otherwise the native module load
  // crashes with a raw "Cannot find module ...onnxruntime_binding.node" that
  // ONNX_WEB_BACKEND=wasm cannot rescue (#1516). HTTP mode was already handled
  // above, so this only blocks the local-runtime path.
  const runtimeBlocker = getLocalEmbeddingRuntimeBlocker();
  if (runtimeBlocker) {
    throw new Error(runtimeBlocker);
  }

  // Return existing instance if available
  if (embedderInstance) {

View on GitHub (pinned to d540b00184)

Solutions

1. Replace initEmbedder() calls with embedText()/embedBatch() — they handle both HTTP and local modes.
2. If you must branch, gate on isHttpMode() from http-client.js and skip init in HTTP mode.
3. Run `gitnexus embeddings status` or check the two env vars to confirm which mode you are in.

Example fix

// before
const pipe = await initEmbedder();
const vec = await embedText(text);
// after — no init needed in either mode
const vec = await embedText(text);

import { isHttpMode } from 'gitnexus/src/core/embeddings/http-client.js';
// Never call initEmbedder() in HTTP mode.
if (!isHttpMode()) {
  await initEmbedder(onProgress);
}

import { isHttpMode } from 'gitnexus/src/core/embeddings/http-client.js';
const shouldUseLocalPipeline = (): boolean => !isHttpMode();

When it happens

Trigger: Any code path that calls initEmbedder() (or the MCP embedder entry point that mirrors it) while GITNEXUS_EMBEDDING_URL and GITNEXUS_EMBEDDING_MODEL are both exported. Typically a custom integration or a forked analyze flow that unconditionally initializes the local pipeline without checking isHttpMode() first.

Common situations: A user sets the two HTTP env vars to route embeddings through an OpenAI-compatible endpoint but has a wrapper script that still calls initEmbedder(); migrating from local to HTTP mode and forgetting to remove the init call; an MCP client that calls both init and embed.

Related errors

• getEmbedder() is not available in HTTP embedding mode. Use e
• Embedder not initialized. Call initEmbedder() first.
• ${name} must be a positive integer, got "${raw}"
• ${name} must be a positive integer <= ${max}, got "${raw}"
• ${name} must be a non-negative integer, got "${raw}"

AI-assisted analysis of abhigyanpatwari/GitNexus@d540b00184 (2026-08-12). Data as JSON: /api/errors/c9857f502134cb6a. Report an issue: GitHub.