Standup

Multi-agent orchestration: Better decisions through collaborative specialist perspectives.

Workflow Routing

Examples

Example 1: Architecture design review

User: "Review this microservices architecture design"
Skill loads: Standup → RunStandup workflow
Process:
  1. Load project context
  2. Present architecture to agents (Product, Security, QA)
  3. Each agent provides perspective
  4. Synthesize recommendations
Output: Design feedback with security, testability, and business concerns identified

Example 2: Feature prioritization

User: "Should we add OAuth2 to MVP or defer to v1.1?"
Skill loads: Standup → RunStandup workflow
Process:
  1. Load project context
  2. Present decision to agents
  3. Product: User value perspective
  4. Security: Security/compliance perspective
  5. QA: Testing complexity perspective
  6. Synthesize: Balanced recommendation
Output: Prioritization decision (Must Have / Should Have / Could Have / Won't Have)

Example 3: Smart roster selection (auto-suggest)

User: "Review this authentication feature design"
Skill loads: Standup → RunStandup workflow
Smart roster: Daniel, Mary, Clay, Hefley, Amy (full team - critical feature)
Process:
  1. Load project context
  2. Auto-detect: authentication = critical feature → suggest full team
  3. Daniel: Security threats (credential storage, session management)
  4. Mary: User experience (signup friction, password reset flow)
  5. Clay: Timeline estimates (6h for email/password, 9h for OAuth2)
  6. Hefley: Business priority (MVP vs v1.1)
  7. Amy: Test requirements (60 tests - 25 unit + 12 integration + 8 E2E + 15 security)
  8. Synthesize: Team consensus with conflict resolution
Output: Multi-perspective design review with actionable recommendations

Example 4: Custom roster - Code review

// Specify exact roster for focused code review
const result = await runStandup({
  feature: 'Database migration script',
  roster: ['Daniel', 'Clay'],  // Only security + tech lead
  codeSnippet: migrationScript
})
// Output: Daniel finds SQL injection risks, Clay estimates timeline

Example 5: Custom roster - Full team override

// Force full team review (override smart defaults)
const result = await runStandup({
  feature: 'Minor UI tweak',
  roster: ['Mary', 'Clay', 'Hefley', 'Daniel', 'Amy'],  // All 5 agents
  description: 'Change button color from blue to green'
})
// Output: Full team perspective on what seems like a simple change
//         (may uncover accessibility, brand, or UX issues)

Example 6: Custom roster - Domain-specific experts

User: "Run standup with Investment Advisory Team to review portfolio allocation strategy"
Skill loads: Standup → RunStandup workflow with custom roster
Agents: Financial Analyst, Compliance Officer, Client Advisor
Output: Investment strategy with risk, compliance, and client perspective

Example 7: Custom roster - Two agents only

// Minimal roster for quick checks
const result = await runStandup({
  feature: 'Performance optimization',
  roster: ['Clay', 'Amy'],  // Only tech lead + QA
  question: 'Is this optimization worth the complexity?'
})
// Output: Clay's capacity estimate + Amy's testing complexity

Example 8: Record decision

User: "Update project context with our decision on auth approach"
Skill loads: Standup → ManageContext workflow
Output: project-context.md updated with decision, rationale, and action items

Agent Rosters

Smart Roster Selection (Auto-Suggest)

FORGE automatically suggests the right experts based on your feature context:

Question Context Override: Questions override feature patterns

• •"How long?" → Clay, Hefley, Amy (timeline focus)
• •"How many tests?" → Amy, Daniel, Clay (testing focus)
• •"Should we build this?" → Hefley, Mary, Clay (prioritization focus)

Manual Override: Explicitly specify roster to override smart defaults

runStandup({ feature: 'Auth', roster: ['Daniel', 'Clay'] }) // Override: only Daniel + Clay

Software Development Roster

• •Daniel (Security Engineer): Security threats, CMMC compliance, secure design
• •Mary (Business Analyst): User value, UX design, user research, stakeholder communication
• •Clay (Tech Lead): Technical feasibility, timeline estimates (Claude-time), capacity planning
• •Hefley (Product Manager): User value, business priorities, MVP scoping, MoSCoW prioritization
• •Amy (QA Lead): Test strategy, testability, quality gates, ATDD

Custom Rosters (Your Domain)

Define your own agent rosters for different domains:

• •Investment Advisory: Financial Analyst, Compliance Officer, Client Advisor
• •Legal Review: Contract Specialist, Risk Manager, Business Counsel
• •Healthcare: Clinical Specialist, Regulatory Affairs, Patient Advocate
• •Product Design: UX Designer, Brand Manager, Accessibility Expert

How to Create Custom Agents: Use templates/custom-agent-template.md to define new agents with:

• •Role, expertise, personality
• •Standup participation style
• •Integration with other agents

Integration

• •Works with AgilePm skill (standup reviews PRDs, prioritizes epics)
• •Works with Security skill (Daniel uses threat modeling in standup)
• •Works with TestArchitect skill (Amy defines test strategy in standup)
• •Generates project-context.md (project "bible" for all agents)
• •Records decisions with rationale (audit trail for compliance)

Methodology

This skill follows multi-agent orchestration principles:

• •Diverse perspectives: Multiple specialists find more issues than solo agent
• •Smart roster selection: Auto-suggest experts based on feature context (authentication → full team, security → Daniel/Clay/Amy)
• •Structured discussion: Each agent speaks in turn, providing their unique expertise
• •Synthesis over voting: Find consensus that's better than any single perspective
• •Decision documentation: Record rationale in project-context.md

Core Innovation: Standup finds 2-3x more issues than solo agent mode (validated Week 8)

Key Features:

• •Context-aware roster suggestion: Automatically suggests the right experts (implemented)
• •Question context override: Questions override feature patterns ("How long?" → planning team)
• •Manual override: Explicitly specify roster to override smart defaults
• •Synthesis with conflict detection: Identifies disagreements and finds middle ground

Based on: Multi-agent systems, Ensemble learning, Scrum standup ceremonies (adapted for AI)

Customization (Release 0.3)

Future enhancements:

• •Configurable rosters: Define agent teams per project type
• •Agent voting: Tie-breaking when consensus fails
• •Historical decision search: "Why did we decide X?" answered from project-context.md
• •Domain-specific roster templates: Pre-defined rosters for finance, healthcare, legal, etc.