Continue not making changes to code

[tylerjborg]

I can make certain changes to the code file, but if I ask Continue to make edits to blocks of code it tries but doesn't succeed. It was working just fine yesterday.

[vldF]

It seems you need to provide more information. A video, for example

[peterHeuz]

I supposedly have the same problem, it never worked though but I am also using continue in combination with VSCode for 2 weeks.

My continue chat is connected to a local LLM in LMStudio. The chat works, the agent (I am using agent mode) can see files and proposes edits to them. When I click "accept" on one of these proposed file changes, the file opens in VSCode, flashes once, but no changes are made to the file. When prompting further, the assistant can see that it didn't change anything, but also fails in the following trials.

I am on Visual Studio Code Version: 1.103.0 (Universal) and have also tested some lower versions. Continue was tested on 1.0.22 as well as the current pre-release

[pavel-cursor]

I have same issue when using qwen3:14b locally(Ollama) in Agent Mode and try to apply changes on the files that it suggested, but no changes make. In same time creating new file is working as indeed.

Software:
Continue: 1.2.0
VS Code: 1.103.2
Ollama: latest

Hardware:
CPU: AMD Ryzen 7 7800X3D
RAM: 48Gb
GPU: NVIDIA 4070 Ti Super 16Gb

config.yml

name: Local Assistant
version: 1.0.0
schema: v1
models:
  - name: Nomic Embed
    provider: ollama
    model: nomic-embed-text:latest
    roles:
      - embed
  - name: Autodetect
    provider: ollama
    model: AUTODETECT
context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase

[lainwir3d]

Same Issue here, sometimes edit works but most of the time it doesn't even though the llm says "I successfully added these lines to the file"

[meichthys]

I'm still seeing this in v 1.2.10 when using the recommended model gpt-oss:20b:

Sometimes it says the edits were completed when they weren't and sometimes it shows "❌ Agent tool use":

Error parsing Ollama response: Error: error parsing tool call: raw='from pathlib import Path, Path ', err=invalid character 'r' in literal false (expecting 'a') {"error":"error parsing tool call: raw='from pathlib import Path, Path\n', err=invalid character 'r' in literal false (expecting 'a')"}

[github-actions]

This issue hasn't been updated in 90 days and will be closed after an additional 10 days without activity. If it's still important, please leave a comment and share any new information that would help us address the issue.

[greaterthanstar]

this is driving me crazy. I would think that applying edits to the current file would be fundamental feature in this extension. instead, it just generates code and expects me to paste it in.
now, I'm had long conversations with Gemini about this, and it says like 1) be careful about makign sure the model has apply in the config, and that is is a model that can handle applys. it suggests deepseek-coder-v2. Also, you are supposed to use Chat mode, not agent mode. But I would really like some clarity from the team about exactly what setup is required to get apply to work! I'm tired of copy/pasting. Github Copilot does this seamlessly. We just need to know the exact config to make it work.

[VeraKowalczuk]

I can attest that a reliable apply would be really helpful. We see the same with our custom endpoint