| title | Project Planning Agents | |||
|---|---|---|---|---|
| description | Agents for requirements gathering, architecture decisions, and security planning | |||
| sidebar_position | 1 | |||
| author | Microsoft | |||
| ms.date | 2026-08-06 | |||
| ms.topic | concept | |||
| keywords |
|
Nine agents and three portable skills support structured project planning across requirements, architecture, backlog management, security, and performance. Each agent follows a guided workflow to produce specific deliverables, from business requirements documents to security assessment plans and reliability targets.
These agents bring structure and consistency to activities that teams often handle ad-hoc:
- Guided workflows walk users through each planning activity step by step, reducing ramp-up time and removing guesswork from unfamiliar processes.
- Every output follows a repeatable template, making documents easier to review, compare, and maintain across projects.
- Architecture decision records and security plans are generated alongside the reasoning that produced them, preserving institutional knowledge.
- Requirements agents persist session state, so multi-day planning efforts pick up where they left off.
- Business analysts, architects, and security engineers share a common toolchain, reducing handoff friction between planning stages.
Tip
Project planning agents work best when invoked early in a project lifecycle. Start with requirements gathering, then move to architecture decisions and security planning as the design matures.
| Agent | Sub-Category | Workflow | Persistence | Key Output |
|---|---|---|---|---|
| BRD Builder | Requirements | 3-phase Q&A | JSON state | Business requirements document |
| PRD Builder | Requirements | 7-phase Q&A | JSON state | Product requirements document |
| Meeting Analyst | Requirements | Transcript analysis | Conversational | Extracted product requirements |
| UX UI Designer | Requirements | JTBD and journey mapping | Conversational | User journeys and accessibility requirements |
| ADR Creator | Architecture | 3-phase Frame/Decide/Govern | JSON state | Architecture decision record |
| System Architecture Reviewer | Architecture | Design trade-off review | Conversational | Architecture assessment |
| Network ISA-95 Planner | Architecture | Layered network planning | Conversational | ISA-95 network plan |
| Backlog Manager | Backlog | Classify and dispatch | Tracking files | Tracker operations across ADO, GitHub, and Jira |
| Functional Planner | Backlog | PRD to hierarchy | Tracking files | Reviewed work-item hierarchy handoff |
| Security Planner | Security | 6-phase STRIDE | JSON state | Security model and backlog |
Security Planner is documented with the security agents and is listed here because it participates in project planning. The three backlog executor subagents are dispatched by Backlog Manager rather than invoked directly, so they are not listed as user-facing agents.
| Skill | Sub-Category | Key Output |
|---|---|---|
| requirements-author | Requirements | Drives the BRD/PRD Builder Q&A workflows |
| architecture-diagrams | Architecture | ASCII/Mermaid diagrams from IaC analysis |
| performance-slo-planner | Performance & Reliability | SLIs, SLOs, load models, test matrices, and reliability backlogs |
The BRD Builder follows a three-phase lifecycle (Discover, Define, Govern) and the PRD Builder follows a seven-phase lifecycle (Assess, Discover, Create, Build, Integrate, Validate, Finalize), both driven by the requirements-author skill. Both agents persist session state as JSON files, supporting pause-and-resume workflows across conversations. Their twin architecture means most concepts transfer between them. Learn one, and the other follows naturally.
Note
BRD and PRD builders share the same underlying workflow engine. Switching between them mid-project requires only a scope adjustment, not a restart.
See the BRD & PRD Builders guide for the shared workflow, feature comparison, and invocation details.
Two agents address architecture documentation from different angles. The ADR Creator uses phase-gated, standards-aligned reasoning to guide users through technical decisions (Frame, Decide, Govern), producing architecture decision records. The architecture-diagrams skill analyzes infrastructure-as-code files and project structure to generate ASCII architecture diagrams directly in conversation.
Tip
Pair the ADR Creator with the architecture-diagrams skill: create an ADR for a design decision, then generate a diagram showing how the chosen approach fits the broader architecture.
- ADR Creator: Guided decision reasoning and documentation
- architecture-diagrams skill: Code-to-diagram generation from IaC analysis
The performance-slo-planner skill turns vague "fast and reliable" expectations into measurable SLIs, SLOs, a load model, a test matrix, and a reliability backlog for production readiness.
Tip
When to use: Run this skill after defining requirements (e.g., after BRD/PRD builders) and before security planning. This ensures production-readiness targets and reliability constraints are baked into the system architecture early.
Note: This skill plans the performance strategy; it does not execute load tests. Hand the generated test matrix to Azure Load Testing tooling for execution.
The Security Planner applies STRIDE-based security model analysis across seven operational buckets to produce standards mappings and dual-format backlog handoff. It detects AI/ML components and recommends RAI Planner dispatch when AI elements are present. The agent uses a six-phase conversational workflow with JSON state persistence for tracking plan progress.
Important
Run security planning after architecture decisions stabilize. Changes to infrastructure or service boundaries may invalidate earlier security models.
See the Security Planning guide for the workflow, operational buckets, and invocation details.
- VS Code with the GitHub Copilot Chat extension installed
- Agent definition files from the
project-planningcollection deployed to.github/agents/ - For Security Planner: agent definition files from the
securitycollection - For BRD/PRD builders: a writable
.copilot-tracking/directory for session state persistence - For diagram generation: the architecture-diagrams skill works with infrastructure-as-code files (Terraform, Bicep, ARM, Kubernetes YAML, or Docker Compose) in the repository
Select any agent using the agent picker in the Copilot Chat pane. Each agent starts its guided workflow automatically.
| Scenario | Agent / Skill | Purpose |
|---|---|---|
| New project kickoff | BRD Builder or PRD Builder | Capture requirements before making architecture decisions |
| Architecture decisions | ADR Creator | Evaluate technology choices, design patterns, or infrastructure approaches |
| Visual documentation | architecture-diagrams skill | Generate ASCII or Mermaid architecture diagrams for onboarding or reviews |
| Performance targets | performance-slo-planner | Define SLIs, SLOs, load models, and reliability backlogs before launch |
| Security review | Security Planner | Assess threats and plan mitigations after architecture decisions stabilize |
For greenfield projects, follow this order to build artifacts that feed into each subsequent step:
- Start with the BRD Builder to capture business context, then the PRD Builder for product-level details.
- Use the ADR Creator to document key design decisions, then the architecture-diagrams skill to visualize the resulting architecture.
- Run the Performance & SLO Planner to set measurable SLI/SLO targets, define load models, and generate a reliability backlog.
- Run the Security Planner once the architecture and performance targets are stable to identify threats and plan mitigations.
- RPI Documentation: Task research, planning, and implementation workflows
- Backlog Management: Work and issue lifecycle management across Azure DevOps, GitHub, and Jira
🤖 Crafted with precision by ✨Copilot following brilliant human instruction, then carefully refined by our team of discerning human reviewers.