Assistant

CLI to generate GitHub issues using an LLM (Ollama/Mistral).

Installation

Quick Install (Recommended)

macOS / Linux:

curl -fsSL https://raw.githubusercontent.com/goudyj/assistant-cli/master/scripts/install.sh | bash

Windows (PowerShell):

irm https://raw.githubusercontent.com/goudyj/assistant-cli/master/scripts/install.ps1 | iex

Install Specific Version

curl -fsSL https://raw.githubusercontent.com/goudyj/assistant-cli/master/scripts/install.sh | bash -s v0.1.0

Build from Source

cargo build --release

Configuration

Configure your settings in ~/.config/assistant.json:

{
  "coding_agent": "claude",
  "ide_command": "cursor",
  "projects": {
    "my-project": {
      "owner": "username",
      "repo": "my-repo",
      "labels": ["bug", "feature", "backend", "frontend"],
      "local_path": "/path/to/my-repo",
      "base_branch": "develop",
      "list_commands": {
        "bugs": ["bug"],
        "frontend": ["frontend", "bug"]
      }
    }
  }
}

Global settings

Field	Description
coding_agent	Agent for dispatch: "claude" or "opencode" (default: "claude")
ide_command	IDE to open worktrees: "code", "cursor", etc. (auto-detected if not set)
auto_format_comments	Auto-format issue comments (default: false)
last_project	Auto-managed: remembers last selected project

Project settings

Field	Description
owner	Repository owner (user or organization)
repo	Repository name
labels	Available labels for issue generation
local_path	Local repository path (required for worktree/dispatch features)
base_branch	Base branch for new branches (auto-detects main/master/develop if not set)
list_commands	Custom filter commands (see below)

Custom filter commands

Define shortcuts to filter issues by labels:

"list_commands": {
  "bugs": ["bug"],
  "urgent": ["bug", "priority:high"]
}

These become available as /bugs, /urgent commands in the TUI.

3. Ollama

Install Ollama and pull the model:

ollama pull mistral:7b

Usage

assistant              # Start the TUI
assistant --project my-project   # Start with a specific project
assistant --logout     # Remove GitHub authentication and exit

GitHub Authentication

On first use, authenticate with GitHub:

〉/login
Starting GitHub authentication...

Open this URL in your browser:
  https://github.com/login/device

And enter the code: ABCD-1234

Waiting for authorization...
Successfully logged in to GitHub!

The token is stored in ~/.config/assistant.json.

Using a Personal Access Token

You can also use a GitHub Personal Access Token instead of the OAuth flow. Add it directly to your config file:

{
  "github_token": "ghp_your_personal_access_token",
  "projects": { ... }
}

To create a token, go to GitHub Settings > Developer settings > Personal access tokens and generate a token with repo scope.

Commands

Press / in the TUI to open the command palette.

Command	Description
/login	Authenticate with GitHub via browser
/logout	Remove GitHub authentication
/repository	Open interactive project selector (alias: /repo)
/agent	Select coding agent (Claude Code or Opencode)
/worktrees	Manage worktrees (view, delete, open in IDE)
/prune	Clean up orphaned worktrees
/help	Show help
/quit	Exit
/<custom>	Custom filter commands defined in config (e.g., /bugs)

Example session

$ assistant

# Select a project with /repo, then browse issues
# Generate new issues by typing a description
# Dispatch issues to a coding agent with Enter
# Create PRs from completed work

Environment variables (optional)

# LLM endpoint (default: http://localhost:11434/api/chat)
LLM_ENDPOINT=http://localhost:11434/api/chat

Project structure

src/
├── agents/           # Coding agent integrations
│   ├── claude.rs     # Claude Code dispatch
│   ├── opencode.rs   # Opencode dispatch
│   ├── worktree.rs   # Git worktree management
│   └── session.rs    # Agent session tracking
├── auth.rs           # OAuth Device Flow
├── config.rs         # JSON configuration
├── github.rs         # GitHub API (octocrab)
├── issues.rs         # Issue generation via LLM
├── llm.rs            # Ollama communication
├── tui.rs            # TUI application
├── tui_events.rs     # Event handling
├── tui_draw.rs       # UI rendering
└── main.rs           # Entry point