Skip to main content

Stage 2: Discovery

Overview

Discovery is where engagements take shape. This stage supports requirement gathering, technical research, business requirements documentation, security planning, and architectural exploration. With 14 assets available, Discovery provides the broadest research toolset in the lifecycle.

When You Enter This Stage

You enter Discovery after completing Stage 1: Setup with a configured environment.

NOTE

Prerequisites: HVE Core installation complete, project repository initialized.

Available Tools

ToolTypeHow to InvokePurpose
task-researcherAgentSelect task-researcher agentResearch best practices and technical topics
brd-builderAgentSelect brd-builder agentCreate business requirements documents
security-plannerAgentSelect security-planner agentGenerate security plans and security models
sssc-plannerAgentSelect sssc-planner agentAssess supply chain security posture against OpenSSF standards
rai-plannerAgentSelect rai-planner agentAssess responsible AI risks and generate RAI plans
gen-data-specAgentSelect gen-data-spec agentGenerate data specifications and schemas
adr-creationAgentSelect adr-creation agentDocument architecture decisions
arch-diagram-builderAgentSelect arch-diagram-builder agentGenerate architecture diagrams
ux-ui-designerAgentSelect ux-ui-designer agentDesign user experience and interface concepts
github-backlog-managerAgentSelect github-backlog-manager agentDiscover and triage existing GitHub issues
memoryAgentSelect memory agentStore research findings for later reference
risk-registerPrompt/risk-registerIdentify and track project risks
task-researchPrompt/task-researchQuick research queries without full agent context
dt-coachAgentSelect dt-coach agentGuide teams through Design Thinking methods for user-centered requirements discovery
experiment-designerAgentSelect experiment-designer agentDesign Minimum Viable Experiments to validate unknowns before committing to implementation

Design Thinking as Pre-Research Methodology

NOTE

Teams can invoke dt-coach during Discovery to run scope conversations (Method 1) and design research (Method 2) before engaging the task-researcher agent. Design Thinking provides structured, empathy-driven research techniques that produce validated problem statements and stakeholder maps, strengthening the foundation for subsequent technical research.

Validating Unknowns with Minimum Viable Experiments

NOTE

When Discovery surfaces unknowns across data, technology, or use cases, invoke experiment-designer to design a Minimum Viable Experiment (MVE) before committing to full implementation. The agent guides you through problem discovery, hypothesis formation, vetting criteria, and experiment planning. MVEs resolve uncertainty early (whether hypotheses are validated or invalidated) the results inform your go/no-go decisions and reduce downstream risk.

Role-Specific Guidance

TPMs lead Discovery, producing BRDs and coordinating research across disciplines. Engineers contribute technical feasibility research. Tech Leads evaluate architecture options. Security Architects drive security model analysis. Data Scientists define data requirements.

UX and UI designers use Discovery-stage tools alongside dt-coach for structured user research. The dt-coach agent provides nine Design Thinking methods, including interview planning, environmental observation, and input synthesis, that complement task-researcher workflows with empathy-driven requirements gathering. See the Design Thinking documentation for method details.

Starter Prompts

Select task-researcher agent:

Research best practices for container orchestration with Kubernetes,
focusing on namespace isolation for multi-tenant environments, resource
quota configuration, and secret management approaches like external
secrets operator vs sealed secrets.

Select brd-builder agent:

Create a business requirements document for the customer onboarding portal.
Target enterprise customers with 500+ seats, with the objective of reducing
onboarding time from 2 weeks to 3 days. Include integration requirements
for existing SSO and billing systems and SOC 2 Type II compliance constraints.

Select security-planner agent:

Generate a security plan for the /api/payments endpoint in our
customer-facing REST API. Scope the plan to authentication via
OAuth 2.0 with Azure AD B2C, PCI DSS compliance for payment
tokenization, and a security model covering injection and broken
access control. Exclude infrastructure and network-level controls.

Select sssc-planner agent:

Assess this repository's supply chain security posture against the OpenSSF Scorecard

Select rai-planner agent:

Assess the responsible AI risks for this project based on the security plan

Stage Outputs and Next Stage

Discovery produces BRDs, research summaries, security plans, data specifications, architecture decision records, supply chain security assessments, and RAI assessments. Transition to Stage 3: Product Definition when the BRD is complete (handoff at docs/brds/). TPMs who have a sufficient BRD can skip directly to Stage 4: Decomposition.

🤖 Crafted with precision by ✨Copilot following brilliant human instruction, then carefully refined by our team of discerning human reviewers.