Getting Started

Install Context Pilot, configure your LLM provider, and run your first AI-powered coding session — all in under 2 minutes.

Prerequisites

• Linux — currently Linux-only. macOS planned. Windows via WSL2.
• Rust toolchain — rustup with stable Rust (1.75+)
• Git — for cloning and git integration
• An LLM API key — from any supported provider (Anthropic recommended)

Installation

Quick install (recommended)

git clone https://github.com/bigmoostache/context-pilot
cd context-pilot
./deploy_local.sh

This compiles from source with cargo build --release and installs the binary to /usr/local/bin/cpilot. Takes 2-3 minutes on first build.

Manual install

git clone https://github.com/bigmoostache/context-pilot
cd context-pilot
cargo build --release
# Binary at: target/release/context-pilot
# Copy it wherever you want:
sudo cp target/release/context-pilot /usr/local/bin/cpilot

Configuration

1. Set your API key

Create a .env file in the Context Pilot directory, or export environment variables:

# Anthropic (recommended)
export ANTHROPIC_API_KEY=sk-ant-...

# Or any of these providers:
export DEEPSEEK_API_KEY=sk-...
export GROK_API_KEY=xai-...
export GROQ_API_KEY=gsk_...
export CLAUDE_CODE_API_KEY=sk-ant-...

2. Optional: Web search & scraping

# Brave Search (for brave_search, brave_llm_context)
export BRAVE_API_KEY=BSA...

# Firecrawl (for firecrawl_scrape, firecrawl_search, firecrawl_map)
export FIRECRAWL_API_KEY=fc-...

3. Optional: GitHub integration

# For gh_execute tool (GitHub CLI)
export GITHUB_TOKEN=ghp_...

Your First Session

1. Navigate to your project

cd ~/your-project
cpilot

Context Pilot creates a .context-pilot/ directory in your project root to store state (memories, logs, todos, conversation history). This directory is automatically gitignored.

2. The TUI layout

You'll see a split-screen terminal interface:

• Left sidebar — context panels (tree, files, memories, logs, etc.) with token counts
• Right panel — currently selected panel content
• Bottom — input field for chatting with the AI
• Top bar — status (LLM provider, branch, cost, token usage)

3. Start talking

Just type naturally. The AI has access to all 58 tools and will use them as needed:

> Explore this codebase and describe the architecture

> Fix the bug in src/parser.rs where it panics on empty input

> Create a callback that runs cargo test on every .rs file change

> Search the web for how to implement a red-black tree in Rust

4. Watch the context

As the AI works, you'll see panels appear in the sidebar — opened files, git status, search results, console output. Each panel shows its token count. The AI manages its own context, opening and closing panels as needed.

Essential Keyboard Shortcuts

Next Steps