Cx Agent Studio

Guide and instructions for using Google Customer Experience Agent Studio (CX Agent Studio). Use when creating conversational agents, writing or structuring i...

Security Scan

This is an instruction-only guide for Google CX Agent Studio; it does not request credentials, install code, or contain hidden endpoints and is consistent with its stated purpose.

Purpose & Capability

The name/description (CX Agent Studio guidance) matches the content: design notes, instruction syntax, callbacks, tools, and evaluation guidance. There are no unrelated requirements (no env vars, binaries, or installs) that would be inconsistent with the stated purpose.

Instruction Scope

SKILL.md and referenced docs stay within builder guidance. They explicitly recommend creating Python tools and callbacks to wrap external APIs and manipulate state — this is expected for an agent-builder but means any code you or your team add (Python tools/callbacks) can run arbitrary logic and access data. The skill itself contains no executable code.

Install Mechanism

No install spec and no code files that would be written to disk. Instruction-only skills are lowest-risk from an installation perspective.

Credentials

No required environment variables, credentials, or config paths are declared. The guide discusses integrating connectors and tools in general terms but does not request secrets itself.

Persistence & Privilege

Skill flags are default (always:false, agent-invocable allowed). It does not request permanent presence or modify other skills or system settings. Autonomous invocation is allowed by platform default and not itself a reason to distrust this instruction-only skill.

Assessment

This skill is a documentation-only guide for building CX Agent Studio agents and appears internally consistent. It does not ask for credentials or install code. However, the guide encourages adding Python tools and callbacks — any Python tool or connector you attach to agents can execute arbitrary code and call external services, so: (1) review and vet any Python code or third-party connectors before enabling them, (2) avoid uploading sensitive documents to RAG/data stores you don't control, (3) enable guardrails, logging, and data-redaction settings if you use third-party integrations, and (4) audit permissions for any connectors (e.g., Salesforce, Google APIs) you configure. If you want a higher assurance, ask the publisher for provenance (homepage, owner identity) or request an explicit list of required connectors and example tool code to review before installation.

Customer Experience Agent Studio (CX Agent Studio) is a minimal code conversational agent builder built on the Agent Development Kit (ADK), representing the evolution of Dialogflow CX.

Core Capabilities

• AI-Augmented Building: Generate agents using Gemini with a 1-2 sentence goal.
• Bi-directional Streaming: Ultra-low latency voice interactions.
• Asynchronous Tool Calling: Maintains natural conversation flow during backend calls.

Quick Actions

1. Generating an Agent with AI

To generate an agent automatically:

• Provide a clear 1-2 sentence goal.
• Optionally provide up to 5 knowledge documents (under 8MB total) like FAQs or tool catalogs. Note: Only works for the root agent and empty agents.

2. Architecture & Design

• Agents: Root (steering) agents orchestrate tasks and delegate to sub-agents. Read references/agents.md.
• Flows: Integrate legacy Dialogflow CX flows for deterministic business logic (auth, sequential validation). Read references/flows.md.
• Variables: Store and retrieve runtime conversation data. Read references/variables.md.

3. Writing Agent Instructions

Agent instructions guide the model's behavior, persona, and tool/agent usage.

• Syntax References:
  • Variables: {variable_name}
  • Tools: {@TOOL: tool_name}
  • Sub-Agents: {@AGENT: Agent Name}
• For complex instructions or recommended XML formatting, read: references/instructions.md
• Best Practices: Start simple, use specific/structured instructions, flat parameter structures. Read references/best-practices.md.

4. Tools & Callbacks

• Tools: Connect your agent to external systems. Wrap complex APIs in Python tools to reduce context overhead. Read references/tools.md.
• Callbacks: Advanced Python hooks (before_agent_callback, after_model_callback, etc.) to control execution, validate states, or inject custom JSON payloads. Read references/callbacks.md.

5. Guardrails & Safety

• Guardrails: Protect against prompt attacks and enforce Responsible AI policies. Read references/guardrails.md.

6. Agent Evaluation

Evaluation ensures agent performance via automated test cases.

• Scenario Test Cases: AI-generated simulated user conversations based on a user goal.
• Golden Test Cases: Specific, ideal conversation paths for regression testing.
• For detailed evaluation metrics, personas, and test case creation, read: references/evaluation.md