Mintplex-Labs/anything-llm · error

e.message

Error message

e.message

What it means

Catch-all handler for POST /api/v1/workspace/:slug/chat in AnythingLLM's Developer API. Any exception thrown while running the chat pipeline (LLM provider call, embedding lookup, vector DB query, attachment processing, telemetry/event logging) is returned as HTTP 500 with the raw exception message in an abort-shaped body {id, type:'abort', textResponse:null, sources:[], close:true, error}. The shape intentionally mimics a streaming abort event so clients can reuse one parser for stream and non-stream endpoints.

Source

Thrown at server/endpoints/api/workspace/index.js:724

          attachments,
          reset,
        });

        await Telemetry.sendTelemetry("sent_chat", {
          LLMSelection:
            workspace.chatProvider ?? process.env.LLM_PROVIDER ?? "openai",
          Embedder: process.env.EMBEDDING_ENGINE || "inherit",
          VectorDbSelection: process.env.VECTOR_DB || "lancedb",
          TTSSelection: process.env.TTS_PROVIDER || "native",
        });
        await EventLogs.logEvent("api_sent_chat", {
          workspaceName: workspace?.name,
          chatModel: workspace?.chatModel || "System Default",
        });
        return response.status(200).json({ ...result });
      } catch (e) {
        console.error(e.message, e);
        response.status(500).json({
          id: uuidv4(),
          type: "abort",
          textResponse: null,
          sources: [],
          close: true,
          error: e.message,
        });
      }
    }
  );

  app.post(
    "/v1/workspace/:slug/stream-chat",
    [validApiKey],
    async (request, response) => {
      /*
   #swagger.tags = ['Workspaces']
   #swagger.description = 'Execute a streamable chat with a workspace'

View on GitHub (pinned to 3aec848f28)

Solutions

1. Read the error field of the 500 body and the server console line (console.error(e.message, e)) - the underlying provider message names the real cause
2. Verify the LLM provider credentials and model in the UI under System Preferences, or check env: LLM_PROVIDER, OPEN_AI_KEY (or the matching provider key), EMBEDDING_ENGINE
3. Send the same prompt from the workspace chat UI; if the UI also fails, the problem is provider config, not the API call
4. If the failure is attachment-related, send documents with Content-Type 'application/anythingllm-document' so they are parsed as text, not images
5. If the embedder or vector DB was changed, re-embed the workspace documents

Example fix

// before
await fetch(`${BASE}/api/v1/workspace/${slug}/chat`, {
  method: 'POST',
  headers: { Authorization: `Bearer ${API_KEY}`, 'Content-Type': 'application/json' },
  body: JSON.stringify({ message: 'Summarize the docs' })
});
// after - surface the server's abort payload instead of throwing a generic Error
const res = await fetch(`${BASE}/api/v1/workspace/${slug}/chat`, { /* same */ });
const data = await res.json();
if (!res.ok || data.type === 'abort')
  throw new Error(`workspace chat failed: ${data.error ?? res.status}`);

const ws = await fetch(`${BASE}/api/v1/workspace/${slug}`, { headers: AUTH }).then(r => r.ok);
if (!ws) throw new Error('workspace missing or unreachable - fix before chat');

function isAbortPayload(data) {
  return typeof data === 'object' && data !== null && data.type === 'abort' && typeof data.error === 'string';
}

try {
  const res = await chat(slug, body);
  const data = await res.json();
  if (isAbortPayload(data)) throw new Error(data.error); // carries the provider's real message
} catch (e) {
  if (/api key|unauthorized|401/i.test(e.message)) fixProviderKey(); else throw e;
}

When it happens

Trigger: POST /api/v1/workspace/:slug/chat where: the configured LLM provider key is invalid/missing (e.g. LLM_PROVIDER=openai with no OPEN_AI_KEY); workspace.chatModel points to a model the provider no longer serves; the embedder was changed without re-embedding; a document attachment is sent with a non 'application/anythingllm-document' mime type and the LLM rejects it; the vector DB (e.g. lancedb) files are corrupted or unreadable.

Common situations: Switching LLM_PROVIDER/env vars without updating the workspace's saved chatModel; Docker container missing provider env vars after an image upgrade; database migration skipped after upgrading AnythingLLM; hitting the API before finishing first-run provider setup in the UI.

Related errors

• e.message
• Workspace ${slug} not found.
• Thread ${threadSlug} not found.
• Message is empty
• ${resolvedMode} is not a valid mode.

AI-assisted analysis of Mintplex-Labs/anything-llm@3aec848f28 (2026-08-18). Data as JSON: /api/errors/5c7e74f17d6e7d61. Report an issue: GitHub.