Before submitting your bug report

• I've tried using the "Ask AI" feature on the Continue docs site to see if the docs have an answer
• I believe this is a bug. I'll try to join the Continue Discord for questions
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

Continue version: v1.2.8
Model: Instinct

Description

In constants.ts you have defined a custom system prompt for Instinct.

However, the system prompt is not automatically injected by Continue (see NextEditProvider.ts).

I would argue the lines in NextEditProvider.ts might be adjusted like this?

    const userPrompt = prompts.find((p) => p.role === "user");
    if (!userPrompt) {
      return undefined;
    }
    const messagesToSend =
      this.endpointType === "fineTuned" || !prompts.some((p) => p.role === "system")
        ? [userPrompt as unknown as ChatMessage]
        : (prompts as unknown as ChatMessage[]);

    const msg: ChatMessage = await llm.chat(messagesToSend, token, {
      stream: false,
    });

To reproduce

Activate Next Edit Autocomplete and check the request body that is sent for each next edit completion. The predefined system prompt is not part of the body.

Log output