Safe Execution
Interactive confirmation for destructive commands with safety checks
Your intelligent terminal companion, making command-line operations intuitive and safe.
$ git clone https://github.com/dheerajcl/Terminal_assistant.git
Cloning repository...
$ cd Terminal_assistant && ./install.sh
Dependencies installed
Via Ollama integration
Multiple API options
Ollama-based private processing
# Install Ollama for local LLM's
$ curl -fsSL https://ollama.com/install.sh | sh
# Get base model
$ ollama pull <model_name>
# Configure local mode
$ shellsage config --mode local
$ shellsage setup
? Select operation mode:
▸ Local (Privacy-first, needs 4GB+ RAM)
API (Faster but requires internet)
? Choose local model:
▸ llama3:8b-instruct-q4_1 (Recommended)
mistral:7b-instruct-v0.3
phi3:mini-128k-instruct
? API provider selection:
▸ Groq
OpenAI
Anthropic
Fireworks
Deepseek
✅ Configuration updated!
$ rm -rf /important-folder
🔎 Potentially dangerous absolute path detected
→ 🛠️ Suggested: rm -rf ./important-folder
$ shellsage ask "show system memory usage"
🔍 Translating request...
→ free -h && top -bn1 | grep 'Mem'
$ shellsage ask "backup all .js files"
📝 Creating backup workflow...
→ mkdir -p ./backups && find . -name '*.js' -exec cp {} ./backups \;
Interactive confirmation for destructive commands with safety checks
Context-aware error analysis and automatic fixes for common issues
Seamless switching between local privacy and cloud performance
Support for Groq, OpenAI, Anthropic, Fireworks, Deepseek
Easy switching between llama3, mistral, phi3 and other models
Local processing option keeps all data on your machine