mirror of
https://github.com/tcsenpai/suggestoor.git
synced 2025-06-04 18:30:19 +00:00
66 lines
1.2 KiB
Markdown
66 lines
1.2 KiB
Markdown
# Suggestoor
|
|
|
|
Suggestoor is a tool that helps you find the right command-line tools for your needs.
|
|
|
|
|
|
|
|
## Important
|
|
|
|
At the first run, it will take a while to embed all the man pages.
|
|
It can take up to 1-2 hours.
|
|
This is a one-time operation.
|
|
|
|
## Prerequisites
|
|
|
|
- ollama server running
|
|
- a GPU for embedding (using a CPU is unfeasible)
|
|
|
|
## Usage
|
|
|
|
### Using uv
|
|
|
|
```bash
|
|
uv run python main.py
|
|
```
|
|
|
|
### Using system python
|
|
|
|
```bash
|
|
python suggestoor.py
|
|
```
|
|
|
|
## Installation
|
|
|
|
### Using uv
|
|
|
|
Just run the script and it will install the dependencies for you.
|
|
|
|
### Using pip
|
|
|
|
```bash
|
|
pip install -r requirements.txt
|
|
```
|
|
|
|
## Configuration
|
|
|
|
You can configure the tool by editing the `.env` file.
|
|
|
|
### OLLAMA_HOST
|
|
|
|
This is the host of the ollama server.
|
|
|
|
### OLLAMA_MODEL
|
|
|
|
This is the model that will be used to generate the response.
|
|
|
|
### EMBEDDING_MODEL
|
|
|
|
This is the model that will be used to embed the man pages.
|
|
|
|
### SIMILARITY_K
|
|
|
|
This is the number of similar chunks to retrieve. Increase this number if you want more context. For a 8gb ram machine, 8 to 12 is a good number.
|
|
|
|
### FOLDERS
|
|
|
|
This is the folders of binaries to search for man pages and embed them. |