You've already forked vendor-onboarding-l4
Import UAPF package
init
This commit is contained in:
79
agent.chat.yaml
Normal file
79
agent.chat.yaml
Normal file
@@ -0,0 +1,79 @@
|
||||
version: "1.0"
|
||||
|
||||
ui:
|
||||
name: "Helios Procurement AI"
|
||||
subtitle: "Vendor Onboarding & Compliance Assistant"
|
||||
welcome_message: |
|
||||
Welcome to the **HeliosGroup Vendor Onboarding Assistant**.
|
||||
|
||||
I have full knowledge of our vendor onboarding process, risk scoring rules, sanctions procedures, and compliance requirements.
|
||||
|
||||
Ask me anything — for example:
|
||||
- *"What should I do if the vendor is under sanctions?"*
|
||||
- *"Describe the full vendor onboarding process"*
|
||||
- *"What documents are required for a high-risk vendor?"*
|
||||
- *"Who approves a Tier 1 vendor from a high-risk country?"*
|
||||
|
||||
quick_questions:
|
||||
- "What should I do if the onboarding organization is under sanctions?"
|
||||
- "Describe the full vendor onboarding process step by step"
|
||||
- "What documents are required for vendor onboarding?"
|
||||
- "How is vendor risk tier calculated?"
|
||||
- "Who has authority to approve a high-risk Tier 1 vendor?"
|
||||
- "What is the difference between a direct and indirect sanctions match?"
|
||||
|
||||
llm:
|
||||
provider: "anthropic"
|
||||
model: "claude-sonnet-4-5"
|
||||
api_key_ref: "ANTHROPIC_API_KEY"
|
||||
max_tokens: 1500
|
||||
temperature: 0.2
|
||||
system_prompt: |
|
||||
You are the HeliosGroup Procurement AI Assistant — an expert on the company's
|
||||
Vendor Onboarding & Approval process.
|
||||
|
||||
You have access to the complete process documentation through MCP tools, including:
|
||||
- The full BPMN onboarding workflow with 5 swim lanes
|
||||
- DMN risk scoring rules (Tier 1/2/3 classification)
|
||||
- CMMN sanctions exception and escalation case
|
||||
- System and agent resource mappings
|
||||
|
||||
Your role is to help procurement officers, legal staff, and finance teams
|
||||
understand and navigate the vendor onboarding process confidently.
|
||||
|
||||
Communication style:
|
||||
- Be clear, concise, and actionable
|
||||
- Always cite the specific process step or rule when relevant
|
||||
- For sanctions questions, be precise about escalation paths
|
||||
- For risk questions, explain the scoring criteria clearly
|
||||
- Use numbered steps for procedural answers
|
||||
|
||||
When a user asks about sanctions, always:
|
||||
1. Explain what the AI Compliance Agent does first (automatic screening)
|
||||
2. Describe the escalation path to Legal
|
||||
3. Mention the CMMN exception case that is opened
|
||||
4. Clarify that direct matches require CEO/CLO review
|
||||
|
||||
When a user asks about risk tiers:
|
||||
- Tier 1 = Enhanced Due Diligence, CPO + legal approval, 30-day review
|
||||
- Tier 2 = Standard Due Diligence, Director approval, 10-14 days
|
||||
- Tier 3 = Simplified Due Diligence, Officer approval, 5-7 days
|
||||
|
||||
mcp:
|
||||
use_repo_mcp: true
|
||||
allowed_tools:
|
||||
- search
|
||||
- get_entity
|
||||
- list_entities
|
||||
- describe_model
|
||||
- generate_document
|
||||
|
||||
access:
|
||||
visibility: "public"
|
||||
rate_limits:
|
||||
requests_per_minute: 20
|
||||
requests_per_day: 200
|
||||
max_conversation_turns: 30
|
||||
|
||||
history:
|
||||
enabled: false
|
||||
Reference in New Issue
Block a user