Skip to content

allanbatista/ai-linux-dev-utilities

BranchesTags

Repository files navigation

Ai Linux Dev Utilities (ab)

Unified CLI utilities for development workflows, powered by LLMs via OpenRouter.

Installation

Quick Install (Recommended)

curl -fsSL https://raw.githubusercontent.com/allanbatista/ai-linux-dev-utilities/master/install.sh | bash

This will:

  • Clone the repository to ~/.local/share/ai-linux-dev-utilities
  • Create a Python virtual environment
  • Install all dependencies
  • Optionally create a symlink at /usr/local/bin/ab
  • Optionally enable bash autocompletion

Install Options

# Install to custom directory
curl -fsSL https://raw.githubusercontent.com/allanbatista/ai-linux-dev-utilities/master/install.sh | bash -s -- -d ~/my-custom-dir

# Skip confirmation prompts (auto-accept all)
curl -fsSL https://raw.githubusercontent.com/allanbatista/ai-linux-dev-utilities/master/install.sh | bash -s -- -y

# Show help
curl -fsSL https://raw.githubusercontent.com/allanbatista/ai-linux-dev-utilities/master/install.sh | bash -s -- -h

Manual Installation

# Clone the repository
git clone https://github.com/allanbatista/ai-linux-dev-utilities.git
cd ai-linux-dev-utilities

# Run the local installer
./install.sh

# Or manually:
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt

# Add to PATH (optional - requires sudo)
sudo ln -s $(pwd)/bin/ab /usr/local/bin/ab

# Enable bash autocompletion (optional)
mkdir -p ~/.local/share/bash-completion/completions
ln -s $(pwd)/completions/ab.bash-completion ~/.local/share/bash-completion/completions/ab

Requirements

  • Python 3.8+
  • git (for .aiignore git root detection)
  • gh CLI (optional, for ab git pr-description -c)

Configuration

API Key Setup

export OPENROUTER_API_KEY="your-api-key-here"

Persistent Configuration

Manage with ab config:

ab config init              # Create default config
ab config show              # View current config
ab config set global.language pt-br    # Change language
ab config set models.default "openai/gpt-4o"  # Change default model

Configuration is stored in ~/.ab/config.json:

{
  "version": "1.0",
  "global": {
    "language": "en",
    "api_base": "https://openrouter.ai/api/v1",
    "api_key_env": "OPENROUTER_API_KEY",
    "timeout_seconds": 300
  },
  "models": {
    "default": "nvidia/nemotron-3-nano-30b-a3b:free",
    "small": "nvidia/nemotron-3-nano-30b-a3b:free",
    "medium": "openai/gpt-5-nano",
    "large": "x-ai/grok-4.1-fast"
  }
}

Command Structure

Commands are organized into categories:

ab <category|command> [arguments...]

# Categories:
ab git <command>        # Git utilities powered by LLM
ab util <command>       # General utilities

# Root commands:
ab prompt               # Send context to LLM (OpenRouter)
ab config               # Manage configuration
ab models               # List available LLM models
ab help                 # Show help

Git Commands (ab git)

Command Description
auto-commit Generate commit messages via LLM
branch-name Generate branch name from task description
changelog Generate changelog from commits
pr-description Generate PR title/description via LLM
resolve-conflict Resolve merge conflicts via LLM
rewrite-history Rewrite commit messages via LLM

Utility Commands (ab util)

Command Description
explain Explain code, errors, or concepts via LLM
gen-script Generate scripts from natural language
passgenerator Secure password generator

Commands

ab models

List and explore available LLM models from OpenRouter API.

ab models [COMMAND] [OPTIONS]

Commands

Command Description
list List available models (default)
info <model-id> Show detailed model information

List Options

Option Description
--free Show only free models
--search TEXT Search by name/description
--context-min N Minimum context length
--modality TYPE Filter by modality (text, image, audio, video)
--sort FIELD Sort by: name, context, price
--limit N Limit results (default: 50)
--json Output as JSON

Examples

# List all models
ab models

# Show only free models
ab models list --free

# Search for Claude models
ab models list --search claude

# Models with 128k+ context
ab models list --context-min 128000

# Vision models sorted by price
ab models list --modality image --sort price

# Get details for specific model
ab models info openai/gpt-4o

# Output as JSON
ab models list --json

ab prompt

Send files and prompts to LLMs via OpenRouter API.

ab prompt [OPTIONS] [PATH...]

Features

  • Automatic binary detection: Binary files are automatically skipped
  • .aiignore support: Use gitignore-style patterns to exclude files
  • Multiple file/directory support: Process entire directories recursively
  • Token management: Control context and response token limits
  • Specialist personas: Use pre-configured expert prompts
  • History tracking: All interactions saved to ~/.ab/history/

Options

Option Short Description Default
--prompt TEXT -p Prompt to send to the LLM. Use - for stdin -
--lang LANG - Output language pt-br
--max-tokens N -n Max tokens for total context 900000
--max-tokens-doc N -nn Max tokens per file 250000
--max-completion-tokens N -m Max tokens for response 16000
--unlimited -u No limit on response tokens false
--model NAME - OpenRouter model name config default
--specialist TYPE -s Persona: dev or rm -
--set-default-model NAME - Persist default model to config -
--only-output - Return only model response (no logs) false
--json - Parse and format response as JSON false
--relative-paths - Show relative paths in output false
--filename-only - Show only filename in output false

Examples

# Simple prompt (no files)
ab prompt -p "What is Python?"

# Analyze a single file
ab prompt script.py -p "Explain this code"

# Process entire directory
ab prompt src/ -p "Review for security issues"

# Use specific model
ab prompt --model "openai/gpt-5" file.py -p "Optimize this"

# Read prompt from stdin
echo "Summarize this code" | ab prompt src/ -p -

# Get JSON response
ab prompt data.json -p "Parse and validate" --only-output --json

# Use developer specialist persona
ab prompt app.py -s dev -p "Refactor this function"

# Unlimited response length
ab prompt -u -p "Write a comprehensive tutorial"

# Set default model
ab prompt --set-default-model "openai/gpt-5-mini"

.aiignore File

Create a .aiignore file to exclude files from processing. Uses gitignore syntax:

# Ignore logs
*.log

# Ignore directories
node_modules/
__pycache__/
.git/
dist/
build/

# Ignore sensitive files
.env
*.secret
credentials.json

# Negation (include this file even if matched above)
!important.log

The tool searches for .aiignore files from the current directory up to the git root, combining all patterns found.

ab git auto-commit

Generate commit messages automatically by analyzing staged changes.

Protected branch detection: When on master or main, auto-commit will suggest creating a feature branch before committing.

ab git auto-commit [OPTIONS]
Option Description
-a Stage all files (git add -A) before committing
-y Skip confirmation prompt
-l LANG Output language (default: en) 
# Generate and confirm commit message
ab git auto-commit

# Stage all and commit without confirmation
ab git auto-commit -a -y

ab git branch-name

Generate branch names from task descriptions using LLM.

ab git branch-name [OPTIONS] "description"
Option Description
-c Create and checkout the branch
-p PREFIX Force a specific prefix (feature, fix, chore, etc.)
-y Skip confirmation when creating
-l LANG Output language (default: en) 
# Suggest branch name
ab git branch-name "fix login bug"
# Output: fix/login-bug

# Create and checkout
ab git branch-name -c "add user authentication"

# Force prefix
ab git branch-name --prefix fix "button alignment"
# Output: fix/button-alignment

# With JIRA ticket
ab git branch-name "JIRA-123: implement payment gateway"
# Output: feature/JIRA-123-implement-payment

ab git changelog

Generate changelog/release notes from commits using LLM.

ab git changelog [OPTIONS] [RANGE]
Option Description
-f FORMAT Output format: markdown, plain, json
-c Group commits by type (feat/fix/chore)
-o FILE Write output to file
-l LANG Output language (default: en) 
# Since last tag to HEAD
ab git changelog

# Between two tags
ab git changelog v1.0.0..v2.0.0

# Last 10 commits in JSON
ab git changelog HEAD~10..HEAD -f json

# Write to file with categories
ab git changelog -c -o CHANGELOG.md

ab git pr-description

Generate pull request title and description by analyzing commits and diff.

ab git pr-description [OPTIONS]
Option Description
-c Create PR via gh CLI after generating
-d Create as draft PR
-b BRANCH Specify base branch
-l LANG Output language (default: en)
-y Skip confirmation prompt 
# Generate PR description
ab git pr-description

# Create PR directly
ab git pr-description -c

# Create draft PR against develop
ab git pr-description -c -d -b develop

# Generate in Portuguese
ab git pr-description -l pt-br

ab git rewrite-history

Rewrite commit messages in git history using LLM.

ab git rewrite-history [OPTIONS] [RANGE]
Option Description
--dry-run, -n Preview changes without applying
--smart LLM decides which commits need rewriting
--force-all Rewrite all commits in range
-y Skip confirmation prompts (batch mode)
-l LANG Output language (default: en)
--backup-branch NAME Custom backup branch name
--skip-merges Skip merge commits
--include-merges Include merge commits

Operation Modes:

  • Interactive: Prompts for each commit
  • Smart: LLM evaluates if message needs rewriting (commits < 5 words are auto-rewritten)
  • Force-all: Rewrites all commits in range

Safety Features:

  • Creates backup branch before changes (backup/pre-rewrite-TIMESTAMP)
  • Warning if commits were already pushed (requires force push)
  • --dry-run mode for full preview
# Interactive mode
ab git rewrite-history

# Preview without changes
ab git rewrite-history --dry-run

# Rewrite last 5 commits
ab git rewrite-history HEAD~5..HEAD

# Batch mode in Portuguese
ab git rewrite-history -y -l pt-br

# Force rewrite all commits
ab git rewrite-history --force-all

# Smart mode (LLM decides)
ab git rewrite-history --smart

ab git resolve-conflict

Analyze and resolve merge conflicts using LLM.

ab git resolve-conflict [OPTIONS] [FILE]
Option Description
-y Auto-apply resolutions without confirmation
--dry-run Preview resolutions without applying
-l LANG Output language (default: en) 
# Interactive mode for all conflicted files
ab git resolve-conflict

# Resolve specific file
ab git resolve-conflict src/app.py

# Preview suggestions only
ab git resolve-conflict --dry-run

# Auto-apply all resolutions
ab git resolve-conflict -y

Features:

  • Detects conflict markers (<<<<<<<, =======, >>>>>>>)
  • Extracts both versions with surrounding context
  • Suggests intelligent merged resolution
  • Option to edit manually before applying

ab util explain

Explain code, errors, or technical concepts using LLM with automatic context gathering.

ab util explain [OPTIONS] [INPUT]
Option Description
-c CONCEPT Explain a technical concept
--history N Include last N lines from bash history
--with-files Include directory listing and referenced files
--context-dir PATH Directory for context gathering
-v Verbose/detailed explanation
-l LANG Output language (default: en) 
# Explain entire file
ab util explain src/app.py

# Explain specific line
ab util explain src/app.py:42

# Explain line range
ab util explain src/app.py:10-50

# Explain error message
ab util explain "error: ECONNREFUSED"

# Explain concept
ab util explain --concept "dependency injection"

# From stdin
echo "stack trace" | ab util explain -

# With bash history context
ab util explain --history 20 "command failed"

# With directory context
ab util explain --with-files "file not found"

ab util gen-script

Generate bash/python scripts from natural language descriptions.

Default behavior: Generates minimal one-liner commands. Use --full for complete scripts with error handling.

ab util gen-script [OPTIONS] "description"
Option Description
--lang LANG Script language: bash, python, sh, perl, ruby, node
--type TYPE Script type: script, cron, oneshot (default: oneshot)
--full Generate complete script with error handling
-o FILE Output file path (auto-enables full mode)
--run Execute the generated script immediately
-l LANG Output language for comments 
# One-liner output (default)
ab util gen-script "list all files larger than 100MB"
# Output: find . -size +100M

# Full script with error handling
ab util gen-script --full "backup database"

# Generate Python one-liner
ab util gen-script --lang python "parse CSV and sum column 3"

# Save to file (auto-full mode)
ab util gen-script -o backup.sh "compress and upload to S3"

# Execute immediately
ab util gen-script --run "show disk usage summary"

# Cron-suitable script
ab util gen-script --type cron "backup database daily"

ab util passgenerator

Generate secure passwords with customizable requirements.

ab util passgenerator LENGTH [OPTIONS]
Option Description
--min-digits N Minimum number of digits
--no-punct Exclude punctuation characters 
# 16-character password
ab util passgenerator 16

# 20 chars with at least 4 digits
ab util passgenerator 20 --min-digits 4

# No punctuation
ab util passgenerator 12 --no-punct

Automatic Model Selection

The ab git auto-commit, ab git pr-description and ab git rewrite-history commands automatically select models based on diff size:

Estimated Tokens Model
≤ 128k nvidia/nemotron-3-nano-30b-a3b:free
≤ 256k openai/gpt-5-nano
> 256k x-ai/grok-4.1-fast

History

All prompt interactions are saved to ~/.ab/history/ with:

  • Timestamp and session ID
  • Model and provider info
  • Token usage and estimated cost
  • Full prompt and response
  • Files processed

View the index at ~/.ab/history/index.json.

License

MIT

About

CLI utilities for dev in linux powerd by LLMs that helps in day to day of development.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages