Skip to content

CXAS SCRAPI

The scripting API for Google CX Agent Studio — build, test, deploy, and maintain conversational agents with confidence.

  • Quick install

    pip install cxas-scrapi

    Python 3.10+ · Apache 2.0 · GitHub

What is CXAS SCRAPI?

Building conversational agents on Google CX Agent Studio is powerful — but the platform's raw API can feel low-level when you want to do something practical, like running a full suite of evaluations, linting your agent configs before a release, or wiring your agent into a CI/CD pipeline. That's exactly what CXAS SCRAPI is for.

CXAS SCRAPI is a high-level Python library and CLI that wraps the official google-cloud-ces client. It gives you friendly Python classes for every CX Agent Studio resource — Apps, Agents, Tools, Guardrails, Sessions, Evaluations, and more — plus a cxas command-line tool that lets you pull and push agent configs, run evaluations, and lint your agents against 60+ best-practice rules. Whether you're working in a Jupyter notebook, writing a Python script, or wiring things together in a GitHub Actions workflow, SCRAPI fits naturally into your workflow.

CXAS SCRAPI Architecture

What can you do with it?

  • Python API

    Pythonic classes for every CX Agent Studio resource. Create, read, update, and delete Apps, Agents, Tools, Guardrails, Sessions, Variables, Versions, and more — all from clean, readable Python code.

    API Reference →

  • CLI

    The cxas command-line tool brings your agent configs down to disk with cxas pull, validates them with cxas lint, and sends them back with cxas push. Everything you need to manage agents without touching the UI.

    CLI Reference →

  • Evaluations

    Five evaluation types — Platform Goldens, Local Simulations, Tool Tests, Callback Tests, and Turn Evals — let you validate your agent's behavior end-to-end, from individual tool calls to full multi-turn conversations.

    Evaluation Guide →

  • Linter

    60+ lint rules check your agent configurations against established best practices before you push. Catch problems like missing tool descriptions, unsafe guardrail configurations, and structural issues early.

    Linting Guide →

  • AI Skills

    The skills system brings AI-powered assistance directly into your development workflow. Skills can build, run, debug, and inspect your agents using the SCRAPI API as their backbone.

    Skills Guide →

  • Insights

    Analyze conversation quality with CCAI Insights scorecards. Import, export, and manage scorecards programmatically or via the CLI.

    Insights Guide →

Where to go next

  • New here? Start with Getting Started

    Install SCRAPI, authenticate with Google Cloud, and run your first command in minutes.

    Get Started →

  • Want the CLI reference?

    Every cxas command, flag, and example — all in one place.

    CLI Reference →

  • Building something in Python?

    Full API docs for every class and method, auto-generated from docstrings.

    API Reference →

  • Looking for a guide?

    Step-by-step walkthroughs for evaluations, linting, CI/CD, branching, and more.

    Guides →