Skip to content

Latest commit

 

History

History
134 lines (85 loc) · 3 KB

getting-started.md

File metadata and controls

134 lines (85 loc) · 3 KB

Getting Started

Welcome to CSL-Core. This guide will get you up and running with the Chimera Specification Language environment, from installation to running your first Agent Guard.

1. Installation

CSL-Core is available via pip. It requires Python 3.10+.

pip install csl-core

To verify the installation and check the version:

cslcore --version

2. Quickstart (The First 5 Minutes)

If you have cloned the repository, we have prepared a quickstart/ directory to help you learn the syntax basics immediately.

Navigate to the quickstart folder:

cd quickstart

Step A: Verify a Policy

Try compiling the "Hello World" policy. This checks for syntax errors and logic consistency using the Z3 engine.

cslcore verify 01_hello_world.csl

Step B: Simulate Runtime Behavior

Now, let's see how the policy behaves against input data without writing any Python code.

** Scenario 1: ** Small external transfer (Should be ALLOWED)

cslcore simulate 01_hello_world.csl --input '{"amount": 50, "destination": "EXTERNAL"}'

** Scenario 2: ** Large external transfer (Should be BLOCKED)

cslcore simulate 01_hello_world.csl --input '{"amount": 500, "destination": "EXTERNAL"}'

You can also explore:

  • 02_age_verification.csl: A simple logic gate for numerical constraints.
  • 03_langchain_template.py: A minimal Python script showing how to load CSL in code.

3. Explore Real-World Examples

Once you understand the basics, check the examples/ directory in the repository root. These represent production-grade use cases.

Core Policies (examples/)

  • agent_tool_guard.csl: Protects an LLM Agent from calling dangerous tools.
  • chimera_banking_case_study.csl: A complex financial policy with VIP limits and risk scoring.
  • dao_treasury_guard.csl: Governance rules for a blockchain DAO.

To run the batched examples and see the output in action:

# Runs the Python runner which loads CSL files and simulates various scenarios
python examples/run_examples.py

4. Integrate with Your Agents (Plugins)

CSL-Core is designed to be a drop-in middleware for your AI framework.

LangChain Integration

We provide a native plugin for LangChain in chimera_core/plugins/langchain.py.

You can see a full implementation in examples/integrations/langchain_agent_demo.py.

5. Write Your Own Policy

Ready to build your own? Create a file named my_policy.csl.

1. Define the Domain:

DOMAIN MyGuard {
  VARIABLES {
    action: {"READ", "WRITE", "DELETE"}
    user_level: 0..5
  }
  
  STATE_CONSTRAINT strict_delete {
    WHEN action == "DELETE"
    THEN user_level >= 4
  }
}

2. Verify it:

cslcore verify my_policy.csl

3. Run the REPL to test edge cases:

cslcore repl my_policy.csl
> {"action": "DELETE", "user_level": 2}
BLOCKED

Next Steps

  • Read the Syntax Specification to master the language.
  • Check the CLI Reference for advanced debugging flags.