llama.vscode.fim

Stripped-down version of the ggml-org/llama.vscode extension, focused exclusively on FIM (Fill-In-the-Middle) auto-completion.

---

Features

• Auto-suggest on input with configurable debounce delay
• Accept a suggestion with Tab
• Accept the first line of a suggestion with Shift + Tab
• Accept the next word with Ctrl/Cmd + Right
• Toggle the suggestion manually by pressing Ctrl + L
• Control max text generation time
• Configure scope of context around the cursor
• Ring context with chunks from open and edited files
• Display performance stats
• Use your local or external model server that supports FIM
• Toggle debug logging on/off via settings

Installation

llama.cpp setup

Open the llama-vscode-fim menu by clicking on the status bar indicator or pressing Ctrl+Shift+M, then select "Install/Upgrade llama.cpp". This will install llama.cpp automatically on Mac and Windows. On Linux, get the latest binaries and add the bin folder to your PATH.

Once llama.cpp is installed, you can select and start a model from the menu.

Below are some details on how to install llama.cpp manually (if you prefer it).

Mac OS

brew install llama.cpp

Windows

winget install llama.cpp

Any other OS

Either use the latest binaries or build llama.cpp from source. For more information on how to run the llama.cpp server, please refer to the Wiki.

llama.cpp settings

Here are recommended settings, depending on the amount of VRAM that you have:

• More than 64GB VRAM:

  llama-server --fim-qwen-30b-default

• More than 16GB VRAM:

  llama-server --fim-qwen-7b-default

• Less than 16GB VRAM:

  llama-server --fim-qwen-3b-default

• Less than 8GB VRAM:

  llama-server --fim-qwen-1.5b-default

CPU-only configs

These are llama-server settings for CPU-only hardware. Note that the quality will be significantly lower:

llama-server \
    -hf ggml-org/Qwen2.5-Coder-1.5B-Q8_0-GGUF \
    --port 8012 -ub 512 -b 512 --ctx-size 0 --cache-reuse 256

llama-server \
    -hf ggml-org/Qwen2.5-Coder-0.5B-Q8_0-GGUF \
    --port 8012 -ub 1024 -b 1024 --ctx-size 0 --cache-reuse 256

You can use any other FIM-compatible model that your system can handle. By default, models downloaded with the -hf flag are stored in:

• Mac OS: ~/Library/Caches/llama.cpp/
• Linux: ~/.cache/lllama.cpp
• Windows: LOCALAPPDATA

Recommended LLMs

The plugin requires FIM-compatible models: HF collection

---

Download the Latest Build

Every push to main/master automatically builds and publishes a ready-to-install .vsix via GitHub Actions.

Download: go to the Releases page and grab llama-vscode-fim-<version>.vsix, then install it:

code --install-extension llama-vscode-fim-0.0.45.vsix

Or from the VS Code UI: Extensions view (Ctrl+Shift+X) → ··· menu → Install from VSIX…

The version in the filename always matches the version field in package.json. Tagged releases (v*) also produce a numbered release with auto-generated changelog notes.

---

Building the Extension

Prerequisites

• Node.js v18 or later
• npm v9 or later
• Visual Studio Code v1.100.0 or later
• TypeScript (npm install -g typescript)
• vsce — the VS Code Extension CLI (npm install -g @vscode/vsce)

1. Clone and install dependencies

git clone https://github.com/arcv/llama.vscode.fim
cd llama.vscode.fim
npm install

2. Compile TypeScript

Compile the extension once:

npm run compile

Or watch for changes during development (recompiles on every file save):

npm run watch

The compiled output lands in the dist/ directory.

3. Run in development mode (Extension Development Host)

The easiest way to test without packaging:

1. Open the repository folder in VS Code:

   code .

2. Press F5 (or go to Run → Start Debugging).

VS Code will launch a new Extension Development Host window with the extension loaded. Any changes you make after recompiling (or while npm run watch is running) take effect after reloading the host window with Ctrl+Shift+P → Developer: Reload Window.

4. Run tests

npm test

This compiles the project and runs the Mocha test suite inside a VS Code test host via @vscode/test-electron.

5. Package as a .vsix file

To produce a self-contained installable package:

vsce package

This generates a file like llama-vscode-fim-0.0.45.vsix in the project root.

Note: vsce will warn if the repository field in package.json points to a non-existent URL or if there are missing fields. Fix any warnings before distributing.

6. Install the .vsix into VS Code

From the command line:

code --install-extension llama-vscode-fim-0.0.45.vsix

From the VS Code UI:

1. Open the Extensions view (Ctrl+Shift+X).
2. Click the ··· menu (top-right of the Extensions panel).
3. Select Install from VSIX…
4. Browse to the generated .vsix file and confirm.

Reload VS Code when prompted. The status bar will show ✓ llama.vscode.fim once the extension activates.

---

Quick Start after Installation

1. Start your llama-server with a Qwen FIM model (see llama.cpp settings above).
2. Open the extension menu (Ctrl+Shift+M) and select Completion models… → Select/start completion model.
3. Pick or add your model endpoint (http://127.0.0.1:8012 by default).
4. Start typing — completions appear automatically. Press Tab to accept.

---

Configuration

Key settings available in VS Code Settings (llama-vscode-fim prefix):

Setting	Default	Description
endpoint	http://127.0.0.1:8012	URL of your llama.cpp server
ai_model	""	Model name for the completion endpoint
debounce_ms	0	Delay (ms) before requesting completion after typing stops
n_prefix	256	Number of prefix lines sent as context
n_suffix	64	Number of suffix lines sent as context
n_predict	128	Max tokens to generate
max_parallel_completions	3	Number of parallel completions (cycle with Alt+] / Alt+[)
debug_log_enabled	true	Enable or disable debug log output
debug_log_level	1 (INFO)	Minimum log level: 0=DEBUG, 1=INFO, 2=WARN, 3=ERROR
debug_log_to_file	true	Write logs to a file in the extension log directory
completion_models_list	[]	List of configured completion models

Keybindings

Shortcut	Action
Tab	Accept suggestion
Shift + Tab	Accept first line of suggestion
Ctrl + Right	Accept first word of suggestion
Ctrl + L	Trigger inline completion manually
Ctrl + Shift + L	Trigger non-cached completion
Alt + ]	Next suggestion
Alt + [	Previous suggestion
Ctrl + Shift + M	Open llama-vscode-fim menu
Ctrl + Shift + ,	Copy chunks (debug)

License

MIT License

Copyright (c) 2026 llama.vscode.fim contributors

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.