发现优质的 AI Agent 技能

Use this skill when you need to migrate a Dialogflow CX agent into CX Agent Studio (Gemini Enterprise for Customer Engagement / CES). It exports the full CX agent package, enumerates all components (intents, entities, flows, pages, webhooks, route groups, etc.), and creates a CES app + agent that runs the CX agent as a **remote Dialogflow agent**.

来自 openclaw/skills 技能

Manage Google Dialogflow CX agents via REST API.

Variables store and retrieve runtime conversation data.

Tools connect agents to external systems or inline code to fetch, update, format, or analyze info.

Agent instructions provide detailed natural-language guidance for the model (goal, persona, tool/sub-agent use).

Guardrails are checks and balances protecting agent applications (model input and output).

Dialogflow CX flows can be integrated into CX Agent Studio agents. - Flow-based agents hand off conversation to the flow until `END_SESSION` is reached, returning to the steering agent. - Agent apps send user input to flows using the `DetectIntent` API. - Require permissions: `Customer Engagement Suite Service Agent` needs `Dialogflow API Client` role.

Evaluation helps automate testing, catch regressions, and measure agent quality via test cases.

Callbacks let you hook into an agent's execution process using Python code to observe, customize, and control behavior at specific points.

This guide provides best practices for designing reliable agent applications.

An agent application is composed of one or more agents: - **Root agent (steering agent)**: Primary entry point and orchestrator. Handles main user interactions, understands overall goals, and delegates tasks. - **Sub-agent (child agent)**: Specialized agent for specific tasks/domains (e.g., searching a database). Promotes modularity.