메뉴
Use when the user wants to build, create, or set up a new Cognigy AI Agent from scratch — covers listing projects, ensuring an LLM exists, creating the agent, testing it, and refining persona/job fields.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Use when adding custom logic inside a Cognigy tool branch with manage_flow_nodes — supported node types, config schemas, placement rules, and the tool-first workflow.
Use when the user wants to add xApps to a Cognigy flow — interactive HTML / Adaptive Card mini-apps rendered to the user, with the submitted data flowing back into the conversation. Covers the xApp flow nodes, the Init Session rule, and accessing xApp data.
Add a new MCP tool (or extend an existing one) in the NiCE Cognigy Plugin's MCP server
Use when the user wants to add knowledge, RAG, or a knowledge store/source to a Cognigy agent — covers embedding vs Knowledge Search models, Knowledge AI settings, ingesting sources, and attaching knowledge as a tool.
Use when configuring or choosing an LLM for a Cognigy agent — valid provider names (openAI, anthropic, azureOpenAI, google, mistral), model strings, connection types, and credential resolution.
Use when exporting, importing, uploading, inspecting, or downloading Cognigy package zip files — including reusing an LLM plus its connection across projects.
| name | agent-creation |
| description | Use when the user wants to build, create, or set up a new Cognigy AI Agent from scratch — covers listing projects, ensuring an LLM exists, creating the agent, testing it, and refining persona/job fields. |
referenceId from the response if you need to assign it explicitly later.
(See the llm-providers skill for valid provider/modelType values)Important: before creating a new tool, inspect the existing agent flow/tools first and make sure you are not reusing an existing toolId. Tool IDs must be unique within an agent flow.
update_ai_agent updates two layers: the AI Agent resource and the AI Agent Job Node.
| Field | Purpose | Example |
|---|---|---|
| name | Agent display name | "Acme Support Bot" |
| description | Agent persona — defines who the agent is, its personality, tone, and high-level behavior. This is the primary field for shaping agent identity. | "You are a friendly customer support agent for Acme Corp. You speak in a professional yet warm tone..." |
| instructions | Agent instructions — high-level guidance and constraints (up to 1000 chars). Use for guardrails and policies that apply regardless of the job. | "Never share internal pricing. Always verify customer identity before account changes." |
| Field | Purpose | Example |
|---|---|---|
| jobName | Job title — the role the agent is performing | "Customer Support Specialist" |
| jobDescription | Job description — detailed description of responsibilities, scope, and expertise areas for this job | "Handle customer inquiries about orders, returns, and product information. Escalate billing disputes to human agents." |
| jobInstructions | Job instructions — specific behavioral rules, step-by-step procedures, and output format requirements for the job | "1. Greet the customer. 2. Ask for their order number. 3. Look up the order using the search tool..." |
| llmProviderReferenceId | Which LLM the agent uses (from setup_llm or list_resources { resourceType: "llm_model" }) | "abc-123-def-456" |
| temperature | LLM temperature (0-1). Lower = more deterministic, higher = more creative. Default: 0.7 | 0.3 |
| maxTokens | Max tokens for LLM response (100-8000). Default: 4000 | 2000 |
update_ai_agent {
aiAgentId: "...",
name: "Acme Support Bot",
description: "You are a friendly and professional customer support agent for Acme Corp. You help customers with orders, returns, and product questions. You maintain a warm but efficient tone.",
instructions: "Never share internal pricing or competitor comparisons. Always verify identity before account changes.",
jobConfig: {
jobName: "Customer Support Specialist",
jobDescription: "Handle customer inquiries about orders, returns, shipping, and product information. You have access to the order database and knowledge base.",
jobInstructions: "1. Greet the customer warmly. 2. Identify their issue. 3. Use the search tool to look up relevant information. 4. Provide a clear answer. 5. Ask if there is anything else you can help with.",
temperature: 0.3,
maxTokens: 2000
}
}
Always use ALL relevant fields when configuring an agent. Do not put everything in description alone — distribute the configuration across the appropriate fields for best results.
toolId values can cause empty or failed responses. Check the flow/tools before assuming the problem is the LLM or connection.