Introducing Whisp - Your AI Co-pilot for the Terminal

We're excited to announce Whisp, an AI-powered shell assistant that brings natural language to your terminal.

Why We Built Whisp

Every developer has experienced that moment — you know what you want to do, but the exact command syntax escapes you. Whether it's a complex find command, arcane git operations, or system administration tasks, memorizing every flag and option is impractical.

Whisp changes this. Describe what you want in plain English, and Whisp translates it into the exact command you need.

Key Features

Natural Language Commands

Type a comma followed by your query:

, find all python files modified in the last week
, show disk usage by directory sorted by size
, list running docker containers with their ports

Whisp generates the command and adds it to your shell history — press Up to use it.

Automatic Error Recovery

When commands fail, Whisp analyzes the error and suggests fixes. No more copying error messages into search engines:

$ npm run build
> Error: Cannot find module 'react'

Whisp suggests:
> npm install react
(Added to history - press Up to use)

Explain and Dry-Run

Understand unfamiliar commands before running them:

,. tar -czvf archive.tar.gz ./docs    # Explain what this does
,d rm -rf ./node_modules              # Preview effects without running

Multiple Providers

Choose from 5 LLM providers based on your needs:

• OpenAI — Fast and reliable (default: gpt-5-nano-2025-08-07)
• Anthropic Claude — Strong reasoning (default: claude-sonnet-4-20250514)
• Ollama — Local and private, no API key needed
• Google Gemini — Large context window
• Cerebras — High-speed inference

Switch providers with a single command:

whisp config set provider ollama
whisp restart

Pipe Mode

Analyze command output with AI:

cat server.log | , summarize errors
git diff | , explain these changes
kubectl get pods | , which pods are failing

Interactive Chat

For complex problems that need back-and-forth:

whisp chat

Safety Features

Whisp protects you from dangerous commands:

• Destructive command detection: Confirms before running rm -rf, dd, mkfs, fork bombs
• Secret redaction: API keys, tokens, and credentials are stripped before sending to LLMs

Metrics and Cost Tracking

Monitor your usage:

whisp metrics              # Response times, token usage
whisp metrics --period day # Filter by time period
whisp health               # Daemon uptime, memory, requests

Architecture

Whisp runs as a lightweight daemon that handles all LLM communication. This means:

• Fast responses (no cold starts)
• Session context across commands
• Automatic rate limiting and retries
• Works offline with Ollama

Shell Integration

Whisp integrates with bash, zsh, and fish. All shortcuts work the same across shells:

Getting Started

Install Whisp in one line:

curl -fsSL https://github.com/n-f9/whisp/releases/download/latest/install.sh | bash

Or build from source:

git clone https://github.com/n-f9/whisp.git
cd whisp
./install.sh

Run the setup wizard:

whisp init

Check out the installation guide for detailed instructions.