# towl 🦉
A fast command-line tool that scans your codebase for TODO comments, lets you browse them in an interactive TUI, and optionally creates GitHub issues from them.
## Features
- **AI Validation**: Use `--ai` to validate TODOs with an LLM (Claude API, OpenAI API, or local CLI agents like Claude Code and Codex) -- filters stale TODOs and enriches GitHub issues
- **Interactive TUI**: Browse, filter, sort, and peek at TODOs in a full-screen terminal interface
- **GitHub Integration**: Create GitHub issues from selected TODOs and replace comments with issue links
- **Smart Detection**: Finds TODO, FIXME, HACK, NOTE, and BUG comments
- **Multiple Output Formats**: JSON, CSV, Markdown, TOML, terminal table (non-interactive mode)
- **Filtering & Sorting**: Filter by TODO type, sort by file, line, type, or priority
- **Fast**: Async I/O, concurrent file scanning, compiled regex patterns
- **Configurable**: Customise file extensions, patterns, and exclusions via `.towl.toml`
- **Context-Aware**: Shows surrounding code lines and enclosing function names
## Installation
```bash
cargo install towl
```
## Quick Start
```bash
# Scan current directory (opens interactive TUI)
towl scan
# Scan in non-interactive mode (CI/scripting)
towl scan -N
# Output to JSON file
towl scan -N -f json -o todos.json
# Filter by type
towl scan -N -t todo
# Create GitHub issues from TODOs
towl scan -N -g
# Preview GitHub issues without creating them
towl scan -N -g -n
# AI-validate TODOs (filters out stale ones)
towl scan -N --ai
# AI + GitHub: create issues for valid TODOs only
towl scan -N --ai -g
# Initialise config
towl init
# Show current config
towl config
```
## Usage
```bash
towl scan [OPTIONS] [PATH]
Options:
-N, --non-interactive Disable interactive TUI mode (for CI/scripting)
-f, --format <FORMAT> Output format (non-interactive only) [default: terminal]
[possible values: table, json, csv, toml, markdown, terminal]
-o, --output <OUTPUT> Output file path (required for json, csv, toml, markdown)
-t, --todo-type <TYPE> Filter by TODO type
[possible values: todo, fixme, hack, note, bug]
-v, --verbose Enable verbose output
-g, --github Create GitHub issues for found TODOs
-n, --dry-run Preview GitHub issues without creating them
--ai Analyse TODOs with AI to validate relevance
towl init [OPTIONS]
Options:
-p, --path <PATH> Config file path [default: .towl.toml]
-F, --force Overwrite existing config file
towl config Show current configuration
```
## Interactive TUI
By default, `towl scan` opens an interactive terminal interface:
| `j` / `Down` | Move cursor down |
| `k` / `Up` | Move cursor up |
| `Space` | Toggle selection |
| `a` | Select all visible |
| `n` | Deselect all |
| `f` | Cycle type filter |
| `s` | Cycle sort field (file, line, type, priority) |
| `r` | Reverse sort order |
| `p` | Peek at source code around the TODO |
| `d` | Delete selected invalid TODOs (with `--ai`) |
| `Enter` | Confirm selection and create GitHub issues |
| `q` / `Esc` | Quit |
Use `--non-interactive` / `-N` to bypass the TUI for CI pipelines and scripting.
## Configuration
Create a `.towl.toml` file in your project root (or run `towl init`):
```toml
[parsing]
file_extensions = ["rs", "toml", "json", "yaml", "yml", "sh", "bash"]
exclude_patterns = ["target/*", ".git/*"]
include_context_lines = 3
[llm]
provider = "claude" # "claude", "openai", "claude-code", or "codex"
model = "claude-opus-4-6"
```
GitHub owner and repo are always auto-detected from `git remote get-url origin` at runtime. Set secrets via environment variables (never stored in the config file):
| `TOWL_GITHUB_TOKEN` | GitHub personal access token |
| `TOWL_LLM_API_KEY` | LLM API key (Claude or OpenAI) |
See the [configuration guide](https://glottologist.github.io/towl/getting-started/configuration.html) for all options.
## License
MIT