Multi-Agent Action System
Purpose
Design a multi-agent operating model for Dataverse and Power Platform actions.
Use this skill to define:
- Sales Agent for quotes, forecasts, pipeline, and customer analysis
- Operations Agent for processes, tickets, and escalations
- Compliance Agent for audits, DLP, policies, and risk analysis
- Data Steward Agent for data quality, duplicates, ownership, and validation
- ALM Agent for deployments, solution checks, and dependency validation
V1 Boundary
This skill defines roles, responsibilities, handoffs, and governance only. It does not spawn or operate live agents.
Workflow
- Identify the business domain and required specialist agents.
- Assign each agent clear responsibilities, allowed decisions, required evidence, and escalation boundaries.
- Define handoff contracts between agents: input, output, confidence, risks, and approvals.
- Define shared memory and audit requirements while avoiding unauthorized cross-domain data exposure.
- Define governance: owner, approval model, monitoring, failure handling, and human override.
Output Format
Return:
Agent map: specialist agents and responsibilities.Handoff model: inputs, outputs, confidence, and escalation rules.Shared context: what can be shared and what must stay restricted.Governance model: approvals, audit, monitoring, and override.Implementation path: skills, flows, queues, and Dataverse structures to design.
1---2name: multi-agent-action-system3description: Use when designing specialized Dataverse AI agents for sales, operations, compliance, data stewardship, ALM, or other enterprise action domains. Produces an operating model only and does not run agents.4---56# Multi-Agent Action System78## Purpose9Design a multi-agent operating model for Dataverse and Power Platform actions.1011Use this skill to define:12- Sales Agent for quotes, forecasts, pipeline, and customer analysis13- Operations Agent for processes, tickets, and escalations14- Compliance Agent for audits, DLP, policies, and risk analysis15- Data Steward Agent for data quality, duplicates, ownership, and validation16- ALM Agent for deployments, solution checks, and dependency validation1718## V1 Boundary19This skill defines roles, responsibilities, handoffs, and governance only. It does not spawn or operate live agents.2021## Workflow221. Identify the business domain and required specialist agents.232. Assign each agent clear responsibilities, allowed decisions, required evidence, and escalation boundaries.243. Define handoff contracts between agents: input, output, confidence, risks, and approvals.254. Define shared memory and audit requirements while avoiding unauthorized cross-domain data exposure.265. Define governance: owner, approval model, monitoring, failure handling, and human override.2728## Output Format29Return:30- `Agent map`: specialist agents and responsibilities.31- `Handoff model`: inputs, outputs, confidence, and escalation rules.32- `Shared context`: what can be shared and what must stay restricted.33- `Governance model`: approvals, audit, monitoring, and override.34- `Implementation path`: skills, flows, queues, and Dataverse structures to design.35
rweisssieker-xp