sigmanticai 0.1.5

SigmanticAI — AI-powered hardware verification from the command line

SigmanticAI

AI-powered hardware verification from the command line.

SigmanticAI generates complete chip verification environments — UVM testbenches, SystemVerilog assertions, coverage models, and RTL — from natural language descriptions. The AI agent runs in the cloud; all tool calls (read, write, compile, simulate) execute locally on your machine via a lightweight WebSocket client.

Installation

pip install sigmanticai

Requires Python 3.10+.

sigmanticai: command not found? Your Python scripts directory may not be on PATH. Try python3 -m sigmanticai as a workaround, or add the path shown in pip's install output to your shell profile:

• macOS/Linux: export PATH="$HOME/.local/bin:$PATH" (add to ~/.zshrc or ~/.bashrc)
• Windows: Add %APPDATA%\Python\PythonXX\Scripts to your system PATH

Quick Start

# Sign in (opens browser)
sigmanticai login

# Interactive mode — chat with the AI agent
sigmanticai

# One-shot generation
sigmanticai generate "Build a UVM testbench for a 32-bit AXI4 FIFO"

# Check your account
sigmanticai whoami
sigmanticai status

What It Does

• Generate verification environments — UVM testbenches, scoreboards, coverage, and assertions from a single prompt
• Design RTL modules — counters, FIFOs, arbiters, interfaces, and more
• Edit existing code — point the agent at your codebase and ask for changes
• Run compilation & simulation — Verilator, Questa, VCS, Xcelium, and other EDA tools (configurable via /tool)
• Iterative debugging — the agent automatically fixes compilation errors and re-runs until tests pass
• Multi-vendor EDA support — configure your toolchain once, agents use the right commands everywhere

How It Works

1. You type a prompt in the CLI
2. Your project context (directory tree, tool config, cache) is sent to a cloud worker
3. The AI agent reasons and plans in the cloud
4. Tool calls (file reads, writes, compilation, simulation) execute locally on your machine
5. Results stream back to the agent for the next reasoning step

Your EDA tools and source files stay on your machine. The cloud handles LLM inference only.

Commands

Command	Description
sigmanticai	Interactive REPL (main mode)
sigmanticai login	Sign in via browser
sigmanticai login --email <e> --password <p>	Headless login (CI/scripts)
sigmanticai logout	Clear stored credentials
sigmanticai whoami	Show current user
sigmanticai status	Show plan, quota, account info
sigmanticai jobs	List recent generation jobs
sigmanticai generate "<prompt>"	One-shot generation

Interactive Commands

Inside the REPL:

Command	Description
/tool	Configure EDA tools (simulator, lint, coverage, formal)
/tool reset	Reset tools to defaults (Verilator)
/tool preset <vendor>	Apply vendor preset (synopsys, cadence, siemens, open_source)
/upload <path>	Push a specific file or directory to the server
/help	Show available commands
exit / quit / q	End session

EDA Tool Support

SigmanticAI supports multiple EDA vendors out of the box:

Category	Supported Tools
Lint	Verilator, Synopsys SpyGlass, Cadence HAL, Siemens Questa Lint
Simulator	Verilator, Synopsys VCS, Cadence Xcelium, Siemens Questa, AMD XSIM
Coverage	Synopsys VCS Coverage, Cadence IMC, Siemens Questa Coverage
Formal	Synopsys VC Formal, Cadence JasperGold, Siemens Questa Formal

Configure your toolchain with /tool — the agent automatically uses the correct commands, flags, and error patterns for your tools.

Requirements

• Python 3.10+
• Internet connection (LLM inference runs in the cloud)
• A SigmanticAI account — sign up at sigmanticai.com

Links

• Website: sigmanticai.com
• Documentation: docs.sigmanticai.com
• Issues: github.com/sigmanticai/sigmanticai-cli/issues