🚀 Agent AI: Multi-Provider Hardware IP Agent System

Supports Claude · GPT-4o · Gemini Platforms
Six Untapped Domains · Ready to Run · Mobile-Friendly

Copyright © 2026 SiliconForge / maomaoati-coder. All Rights Reserved.
Commercial use requires a separate license. See LICENSE

---

​OFFICIAL ATTRIBUTION NOTICE This document and the associated source code in this repository are anchored by the following SHA256 Proof of Existence:

05dc346326c70d59ba94e4da7c162b000b7a3fbab396acf54b5f4b791af7c0bf

Unauthorized modification of this file constitutes a breach of the Agent AI Commercial License

📂 Table of Contents

1. What Is This?
2. Supported AI Providers
3. Six Agent Modules
4. Prerequisites
5. Installation
6. Configure Your API Key
7. How to Run
8. Module Reference
9. Troubleshooting
10. Mobile Guide (Android Termux)

---

1. What Is This?

Agent AI is a hardware IP agent system built specifically for independent chip architects. Unlike general-purpose AI tools that focus on generic coding, Agent AI covers six specialized domains in hardware design and IP management.

It provides Physical Sovereignty: you are not locked into one AI provider. Switch between Claude, GPT-4o, and Gemini seamlessly within a single framework.

---

2. Supported AI Providers

Provider	Core Model	Parameter	Env Variable
Anthropic Claude	claude-sonnet-4	claude	ANTHROPIC_API_KEY
OpenAI	gpt-4o	openai	OPENAI_API_KEY
Google Gemini	gemini-3.0-pro	gemini	GEMINI_API_KEY

Important

Live Search Feature: The FA (Fab Intelligence) module uses live web search only when using the Claude provider. OpenAI and Gemini will answer from their internal training data.

---

3. Six Agent Modules

Select a module by starting your input with the first two uppercase letters:

Prefix	Module	Primary Function
IP	IP Lifecycle Orchestration	Automates NDA ➔ Report ➔ Negotiate ➔ Contract ➔ Deliver
HV	Hardware Verification	RTL analysis + Auto-generation of Verilog testbenches
EF	Energy Sovereignty	$EFF = E_{recycled}/E_{total}$ grading of physical architectures
CT	Adversarial Contract Audit	Scans contracts for ownership traps and revenue leaks
FA	Fab Intelligence	PDK selection, tapeout platform intelligence, and cost modeling
TP	Trust Provenance Chain	SHA256 local anchoring + IP ownership certificate generation

---

4. Prerequisites

• Python 3.8+: Essential runtime environment.
• API Key: At least one key from Claude, OpenAI, or Gemini.
• Network: Required for AI API calls and FA module live searches.

Check Python version:

python3 --version  # Must be 3.8.x or higher

5. Installation

Step 1: Clone the Repository

git clone [https://github.com/maomaoati-coder/Agent-AI.git](https://github.com/maomaoati-coder/Agent-AI.git)
cd Agent-AI

Step 2: Install SDKs

Install only the SDKs you have keys for, or install everything:

# Individual installs
pip install anthropic>=0.49.0           # For Claude
pip install openai>=1.30.0              # For OpenAI
pip install google-generativeai>=0.8.0  # For Gemini

# Or install all dependencies
pip install -r requirements.txt

Step 3: Verify Installation

python3 -c "import anthropic; print('Claude SDK OK')"

6. Configure Your API Key

Temporary (Current session):

export ANTHROPIC_API_KEY="sk-ant-your-key-here"

Permanent (Recommended):

echo 'export ANTHROPIC_API_KEY="sk-ant-your-key-here"' >> ~/.bashrc
source ~/.bashrc

7. How to Run

A. Run via CLI Demo

python3 agent_ai_core.py claude    # Run using Claude
python3 agent_ai_core.py openai    # Run using OpenAI
python3 agent_ai_core.py gemini    # Run using Gemini

B. Integrate in Python

from agent_ai_core import AgentAI

agent = AgentAI(provider="claude")
result = agent.run("IP A buyer emailed, how do we start?")
print(result["result"]["draft"])

8. Module Reference

🛡️ IP — Lifecycle Orchestration

Guides your deal stage: NDA ➔ REPORT ➔ NEGOTIATE ➔ CONTRACT ➔ DELIVER.

• Example: IP NDA is signed, they want technical materials.

🏗️ HV — Hardware Verification

Generates EDA-compatible Verilog. Targets: ≥95% code coverage, ≥90% branch coverage.

• Example: HV Generate testbench for 6-state anomaly_fsm.

⚡ EF — Energy Sovereignty

Grade	EFF Range	Meaning
SOVEREIGN	>90%	Energy sovereignty achieved
MID	60–90%	Near target, minor optimization needed
CRITICAL	<30%	Architecture needs redesign

🔍 CT — Adversarial Contract Audit

Scans for "Ownership Traps" (e.g., "Party A owns all IP created during...").

🌐 FA & TP — Fab & Trust

• FA: Node recommendation & cost modeling (e.g., FA SKY130 tapeout cost 2026).
• TP: Local SHA256 anchoring. Anchors can be submitted to Zenodo or GitHub for proof of existence.

9. Troubleshooting

Warning

• AuthenticationError (401): Key is missing or invalid. Check your export commands.
• ModuleNotFoundError: Run the pip install commands from Section 5.
• EnvironmentError: Ensure you have source ~/.bashrc after editing your profile.

10. Mobile Guide (Android Termux)

1. Install Termux: Download from F-Droid (not Google Play).
2. Setup Environment:

pkg update && pkg upgrade
pkg install python git

3. Clone & Run:

git clone [https://github.com/maomaoati-coder/Agent-AI.git](https://github.com/maomaoati-coder/Agent-AI.git)
cd Agent-AI
pip install anthropic
python agent_ai_core.py

💡 Quick Reference Card

Prefix	Function	Example Input
IP	IP Sales	IP A buyer emailed, how do we start?
HV	Verification	HV Generate testbench for anomaly_fsm
EF	Energy	EF Erecycled=85W Etotal=100W
CT	Contract	CT [paste contract text here...]
FA	Fab Intel	FA SKY130 tapeout cost estimate 2026
TP	Provenance	TP SoC V3.0 IP ownership anchor

Agent AI · © 2026 SiliconForge / maomaoati-coder

For commercial inquiries or acquisition: GitHub Repository

---

🔒 Project Authenticity & Provenance

The integrity of this release and its associated Commercial License (v1.0) is cryptographically anchored:

• Fingerprint (SHA256): 05dc346326c70d59ba94e4da7c162b000b7a3fbab396acf54b5f4b791af7c0bf
• Note: This hash represents the immutable logic state of Agent AI as of April 2026. Any unauthorized tampering will invalidate this fingerprint.