PFS - Please Find Solution

PFS is a command-line tool that diagnoses and corrects your last failed shell command using a local or hosted Large Language Model (LLM) powered by Ollama.

When you run a command that results in an error (i.e., has a non-zero exit code), simply type pfs. The tool will analyze the failed command and provide an explanation of the error and a corrected command. You can then choose to execute the corrected command instantly.

Features

• Local First: Works entirely with your local Ollama models, ensuring privacy and offline capability.
• Intelligent Correction: Leverages LLMs to understand the context of your error and provide accurate fixes.
• Interactive Workflow: Explains the error and prompts for confirmation before executing any command.
• Reliable Capture: A robust shell function ensures that the command, its output, and exit code are all captured correctly for analysis.

Requirements

• Go 1.22+: For building the application.
• jq: For safely handling command output in the shell wrapper. You can install it with your system's package manager (e.g., sudo dnf install jq, sudo apt-get install jq, brew install jq).
• ollama: The LLM that application talks with. After installing, ollama must serving (ollama serve &) and at least 1 model is downloaded to your machine (ollama pull deepseek-r1:1.5b).

Installation

1. Clone the Repository

git clone https://github.com/tatlilimon/PFS.git
cd PFS

2. Build the Binary

Build the binary and move the binary to a directory in your system's PATH.

go build -o /usr/local/bin/pfs ./cmd/

3. Install the Binary

sudo mv pfs /usr/local/bin/

Configuration

PFS uses a .env file in your home directory to store your Ollama settings.

1. Create the Configuration File

Copy the provided template to ~/.pfs.env.

cp .env.template ~/.pfs.env

2. Edit the Configuration

Open the file and set the OLLAMA_BASE_URL for your Ollama instance and the OLLAMA_MODEL you wish to use.

nano ~/.pfs.env

3. Choosing a Model

The quality of the command correction depends heavily on the model you choose. For best results, use a model that is specifically fine-tuned for code or command-line instructions.

Shell Setup

To make PFS work, you need to source the provided wrapper script in your shell's startup file (e.g., .bashrc, .zshrc).

1. Add the Wrapper to Your Shell Configuration

Open your shell's configuration file:

# For Bash
nano ~/.bashrc

# For Zsh
nano ~/.zshrc

Add the following line to the end of the file. Make sure to use the actual absolute path to the cloned repository.

source /path/to/your/PFS/pfs_wrapper.sh

2. Reload Your Shell

For the changes to take effect, either restart your terminal or source your configuration file:

# For Bash
source ~/.bashrc

# For Zsh
source ~/.zshrc

How to Use

1. Run any shell command.
2. If it fails, simply type pfs and press Enter.
3. The tool will provide an explanation and a corrected command.
4. Press y and Enter to execute the new command, or n to abort.

Debugging

PFS provides a --verbose flag for debugging purposes, which provides detailed output about the interaction with the Ollama model.

Example:

pfs --verbose

Feel Free to Contribute This Project!

You can help me to developing this app by opening a pull request or issue.