TealTiger

Developer-First AI Security & Cost Governance SDK

Secure your AI. Control your costs. Zero infrastructure.

Drop-in SDKs that add security guardrails, policy enforcement, and cost tracking to your AI applications. Works with OpenAI, Anthropic, and more.

📖 Read the introduction blog post to learn more about TealTiger!

---

✨ What is TealTiger?

TealTiger is an SDK-only AI security and governance platform that provides:

• 🛡️ Policy Enforcement - ENFORCE / MONITOR / REPORT_ONLY modes for safe rollouts
• 🔒 Security Guardrails - PII detection, prompt injection prevention, content moderation
• 💰 Cost Tracking - Monitor and control AI spending across providers
• 📊 Audit Logging - Redaction-by-default audit trails with correlation IDs
• 🎯 Deterministic Decisions - Stable, typed Decision contract with reason codes
• ⚡ Zero Infrastructure - No servers, no deployment complexity

---

🚀 Quick Start

Installation

TypeScript/JavaScript:

npm install tealtiger

Python:

pip install tealtiger

Basic Usage

TypeScript:

import { TealEngine, PolicyMode } from 'tealtiger';

const engine = new TealEngine({
  policies: {
    tools: {
      web_search: { allowed: true },
      file_delete: { allowed: false }
    }
  },
  mode: PolicyMode.ENFORCE
});

const decision = engine.evaluate({
  action: 'tool.execute',
  tool_name: 'web_search',
  context: { user_id: 'user-123' }
});

console.log(decision.action); // 'allow' or 'deny'
console.log(decision.reason_code); // e.g., 'policy.tool.allowed'

Python:

from tealtiger import TealEngine, PolicyMode

engine = TealEngine(
    policies={
        "tools": {
            "web_search": {"allowed": True},
            "file_delete": {"allowed": False}
        }
    },
    mode=PolicyMode.ENFORCE
)

decision = engine.evaluate({
    "action": "tool.execute",
    "tool_name": "web_search",
    "context": {"user_id": "user-123"}
})

print(decision.action)  # 'allow' or 'deny'
print(decision.reason_code)  # e.g., 'policy.tool.allowed'

---

📦 SDK Repositories

TealTiger provides separate repositories for each SDK with full documentation:

SDK	Repository	Package	Documentation
TypeScript	tealtiger-typescript-prod	npm	Docs
Python	tealtiger-python-prod	PyPI	Docs

---

🌟 Key Features

Policy Rollout Modes

Deploy policies safely with three modes:

• ENFORCE: Block violating actions in production
• MONITOR: Log violations without blocking (shadow mode)
• REPORT_ONLY: Collect data for policy tuning

Core Components (v1.1.0)

• TealEngine - Policy evaluation engine with deterministic decisions
• TealGuard - Security guardrails (PII, prompt injection, content moderation)
• TealCircuit - Circuit breaker for cascading failure prevention
• TealAudit - Audit logging with redaction-by-default
• TealMonitor - Performance and cost monitoring

Enterprise Features (v1.1.x)

• Decision Contract - Stable, typed Decision object with action, reason codes, risk scores
• Execution Context - Correlation IDs and traceability across components
• Audit Redaction - Security-by-default with configurable redaction levels
• Policy Testing - CLI and library test runner for CI/CD integration

Provider Support

Current (v1.1.0):

• ✅ OpenAI (GPT-4, GPT-3.5)
• ✅ Anthropic (Claude)

Coming in v1.1.x:

• 🔜 Google Gemini
• 🔜 AWS Bedrock
• 🔜 Azure OpenAI
• 🔜 Cohere
• 🔜 Mistral AI

---

📚 Documentation

• Official Documentation - Complete guides and API reference
• Quickstart Guide - Get started in 5 minutes
• Core Concepts - Understand the decision model
• API Reference - TypeScript - TypeScript API docs
• API Reference - Python - Python API docs
• Migration Guide - Upgrade to v1.1.x
• FAQ - Frequently asked questions

---

🛡️ OWASP Coverage

TealTiger v1.1.0 provides comprehensive coverage for 7 out of 10 OWASP Top 10 for Agentic Applications (ASI01-ASI10) vulnerabilities through its SDK-only architecture.

ASI	Vulnerability	Coverage
ASI01	Goal Hijacking & Prompt Injection	🟡 Partial
ASI02	Tool Misuse & Unauthorized Actions	🟢 Full
ASI03	Identity & Access Control Failures	🟢 Full
ASI04	Supply Chain Vulnerabilities	🔧 Support
ASI05	Unsafe Code Execution	🟢 Full
ASI06	Memory & Context Corruption	🟢 Full
ASI07	Inter-Agent Communication Security	❌ Platform
ASI08	Cascading Failures & Resource Exhaustion	🟢 Full
ASI09	Harmful Content Generation	🔧 Support
ASI10	Rogue Agent Behavior	🟢 Full

Total Coverage: 7/10 ASIs (70%) with SDK alone

Learn more: OWASP ASI Mapping

---

🎯 Use Cases

• Enterprise AI Applications - Policy enforcement and compliance
• Customer Support Bots - Protect customer PII and sensitive data
• Healthcare AI - HIPAA compliance and audit trails
• Financial Services - Prevent data leakage and unauthorized actions
• E-commerce Agents - Secure payment information and transactions
• Education Platforms - Content safety and moderation

---

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

Development Setup

TypeScript SDK:

git clone https://github.com/agentguard-ai/tealtiger-typescript-prod.git
cd tealtiger-typescript-prod
npm install
npm test

Python SDK:

git clone https://github.com/agentguard-ai/tealtiger-python-prod.git
cd tealtiger-python-prod
pip install -e ".[dev]"
pytest

---

📄 License

Apache 2.0 © TealTiger Team

TealTiger SDKs are open source and will always remain free. The future SaaS platform will be a commercial offering.

See LICENSE for details.

---

🔒 Security

Security is our top priority. If you discover a security vulnerability, please see our Security Policy.

---

📞 Support & Community

• Documentation: docs.tealtiger.ai
• GitHub Issues: Report bugs and request features
• Email: reachout@tealtiger.ai
• Blog: blogs.tealtiger.ai

---

🌟 Links

• Website: tealtiger.ai
• Documentation: docs.tealtiger.ai
• Playground: playground.tealtiger.ai
• NPM Package: npmjs.com/package/tealtiger
• PyPI Package: pypi.org/project/tealtiger
• TypeScript SDK: github.com/agentguard-ai/tealtiger-typescript-prod
• Python SDK: github.com/agentguard-ai/tealtiger-python-prod

---

Made with ❤️ for the AI community

⭐ Star us on GitHub | 📖 Read the Docs | 🐦 Follow us on Twitter