You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A new multi-agent workflow where Morgan, Riley, and Sam collaboratively design and plan tabletop incident exercises that combine SRE response (Morgan), infrastructure failover (Riley), and security incident handling (Sam). Produces scenario-based exercise plans, evaluation criteria, and post-exercise review templates. Reads existing BGreat artifacts to generate realistic, cross-domain incident scenarios tailored to the team's actual architecture.
Market Signal
86-89% of enterprise multi-agent pilots fail to reach production at scale (Gartner/industry reports). Tabletop exercises and incident simulation are increasingly recognized as essential for production readiness. Incident simulation tools (Cutover AI-enabled runbooks, PagerDuty, incident.io) focus on runtime execution, not planning-phase scenario design. No planning framework currently generates multi-domain incident scenarios from existing ops artifacts. The BMAD ecosystem's structured multi-agent approach (49K stars, growing adoption) is uniquely positioned to deliver this. Sources: Gartner AI agent adoption reports, Kanerika multi-agent orchestration analysis, Beam.ai production patterns guide.
User Signal
Enhancement issue #60 ("feat: Add multi-agent operations party mode for cross-domain review") has been open since April 2026, demonstrating direct user demand for cross-agent collaboration workflows. The existing ops review skill (OR) checks artifact consistency but does not simulate incidents or generate exercise plans. The gap between having plans and testing those plans is where production failures happen.
Technical Opportunity
Three specialized agents (Morgan/SRE, Riley/DevOps, Sam/Security) already have deep domain expertise documented in their SKILL.md files, and the ops review skill already demonstrates cross-agent artifact reading. A new workflow (e.g., bgr-3-create-incident-simulation, menu code IS) follows the standard pattern:
step-01 reads existing BGreat artifacts to build a scenario inventory (what can fail, what plans exist)
step-02 designs scenario details with each agent contributing their domain perspective (Morgan: SRE response, Riley: infra failover, Sam: security incident handling)
step-03 defines evaluation criteria and success metrics per scenario
step-04 produces exercise templates, facilitator guides, and post-exercise review checklists
step-05 validates completeness and cross-references against existing plans
Naturally becomes the "capstone" workflow after multiple planning workflows are complete, positioned alongside the Operations Review in the recommended workflow order.
Assessment
Dimension
Score
Rationale
Feasibility
high
Content-only workflow following established patterns; uses existing agent expertise and artifact reading already proven in ops review
Impact
high
Unique multi-agent value that no competitor offers; validates all other planning workflows; demonstrates BGreat's cross-domain collaboration capability
Urgency
med
Enhancement #60 has been open for 3+ months; growing user interest but not time-critical like regulatory compliance
Adversarial Review
Strongest objection: If teams haven't completed multiple BGreat workflows, there are insufficient artifacts to cross-reference for realistic scenarios. This workflow only works late in the adoption cycle, limiting its audience.
Rebuttal: The workflow can gracefully degrade — with even one or two completed artifacts, it generates scenarios focused on those domains (e.g., observability + incident response → alert-based scenarios). The recommended workflow order already places Operations Review (OR) after multiple workflows; this simulation workflow fits in the same late-stage position. Moreover, the multi-agent collaboration is itself a demonstration of BGreat's unique value proposition as a multi-agent planning framework, potentially driving adoption of individual workflows by showing teams what a complete readiness posture looks like.
Suggested Next Step
Create a new workflow directory (bgr-3-create-incident-simulation) with the standard SKILL.md + manifest + workflow.md + steps/ + templates/ structure. Register in module-help.csv with menu code IS and after-dependency on at least one existing workflow (e.g., bgr-3-create-incident-response).
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Summary
A new multi-agent workflow where Morgan, Riley, and Sam collaboratively design and plan tabletop incident exercises that combine SRE response (Morgan), infrastructure failover (Riley), and security incident handling (Sam). Produces scenario-based exercise plans, evaluation criteria, and post-exercise review templates. Reads existing BGreat artifacts to generate realistic, cross-domain incident scenarios tailored to the team's actual architecture.
Market Signal
86-89% of enterprise multi-agent pilots fail to reach production at scale (Gartner/industry reports). Tabletop exercises and incident simulation are increasingly recognized as essential for production readiness. Incident simulation tools (Cutover AI-enabled runbooks, PagerDuty, incident.io) focus on runtime execution, not planning-phase scenario design. No planning framework currently generates multi-domain incident scenarios from existing ops artifacts. The BMAD ecosystem's structured multi-agent approach (49K stars, growing adoption) is uniquely positioned to deliver this. Sources: Gartner AI agent adoption reports, Kanerika multi-agent orchestration analysis, Beam.ai production patterns guide.
User Signal
Enhancement issue #60 ("feat: Add multi-agent operations party mode for cross-domain review") has been open since April 2026, demonstrating direct user demand for cross-agent collaboration workflows. The existing ops review skill (OR) checks artifact consistency but does not simulate incidents or generate exercise plans. The gap between having plans and testing those plans is where production failures happen.
Technical Opportunity
Three specialized agents (Morgan/SRE, Riley/DevOps, Sam/Security) already have deep domain expertise documented in their SKILL.md files, and the ops review skill already demonstrates cross-agent artifact reading. A new workflow (e.g.,
bgr-3-create-incident-simulation, menu codeIS) follows the standard pattern:Naturally becomes the "capstone" workflow after multiple planning workflows are complete, positioned alongside the Operations Review in the recommended workflow order.
Assessment
Adversarial Review
Strongest objection: If teams haven't completed multiple BGreat workflows, there are insufficient artifacts to cross-reference for realistic scenarios. This workflow only works late in the adoption cycle, limiting its audience.
Rebuttal: The workflow can gracefully degrade — with even one or two completed artifacts, it generates scenarios focused on those domains (e.g., observability + incident response → alert-based scenarios). The recommended workflow order already places Operations Review (OR) after multiple workflows; this simulation workflow fits in the same late-stage position. Moreover, the multi-agent collaboration is itself a demonstration of BGreat's unique value proposition as a multi-agent planning framework, potentially driving adoption of individual workflows by showing teams what a complete readiness posture looks like.
Suggested Next Step
Create a new workflow directory (
bgr-3-create-incident-simulation) with the standard SKILL.md + manifest + workflow.md + steps/ + templates/ structure. Register in module-help.csv with menu codeISand after-dependency on at least one existing workflow (e.g.,bgr-3-create-incident-response).All reactions