bmad-observability-agent

B-MAD Agent Integration

The O11y Engineer collaborates with the default B-MAD agents. This page documents how each agent interacts with observability workflows, what artifacts are exchanged, and when handoffs occur.

Interaction Overview

graph TB
    O11y["O11y Engineer"]

    Bob["Bob (Scrum Master)"]
    Amelia["Amelia (Dev)"]
    Murat["Murat (Test Architect)"]
    Winston["Winston (Architect)"]
    John["John (Product Manager)"]
    Mary["Mary (Business Analyst)"]

    O11y -->|companion epics + stories| Bob
    O11y -->|fix stories from validation| Bob
    Bob -->|sprint planning with o11y work| Amelia
    O11y -->|spec_reference in stories| Amelia
    O11y -->|SLO contract| Murat
    O11y -->|collector architecture questions| Winston
    O11y -->|PRD for observability changes| John
    O11y -->|observability requirements| Mary

    style O11y fill:#2563eb,color:#fff
    style Bob fill:#7c3aed,color:#fff
    style Amelia fill:#059669,color:#fff
    style Murat fill:#d97706,color:#fff
    style Winston fill:#dc2626,color:#fff
    style John fill:#0891b2,color:#fff
    style Mary fill:#be185d,color:#fff

Bob (Scrum Master)

Role: Receives companion epics and stories, plans sprints that include observability work, routes fix stories from validation failures.

Artifacts Sent to Bob

Artifact Trigger Location
Companion epic with stories generate-observability-spec _bmad-output/epics/
Fix stories from validation validate-traces Appended to existing epic
Observability change epic plan-observability-change _bmad-output/epics/

How Bob Uses These

!!! note All stories use the standard BMAD story format. Bob does not need any special handling.


Amelia (Dev)

Role: Implements instrumentation stories, follows attribute ownership maps, validates locally.

Artifacts Provided to Amelia

Artifact Purpose Key Fields
Instrumentation stories Implementation tasks spec_reference, test_criteria
Observability spec Reference document Trace/log/metric contracts
Attribute ownership map What to instrument where app_managed vs collector_managed

How Amelia Uses These

spec_reference: Links to the exact section of the observability spec. Amelia uses this to understand which spans to create, which logs to emit, and which metrics to record.

Attribute ownership: Three categories:

!!! tip Always check the attribute_ownership section before instrumenting. Setting collector-managed attributes in application code creates duplicates.


Murat (Test Architect)

Role: Receives SLO contracts for test design, implements assertions matching SLO targets.

Artifacts Provided to Murat

Artifact Purpose Location
SLO contract Test design targets observability-specs/slo-contract.yaml
Test type mapping Which test per KPI Embedded in SLO contract
Validation reports Coverage gaps _bmad-output/o11y-artifacts/reports/

SLO-to-Test Mapping

Each SLO entry includes a test_assertion field:

test_assertion: "p95_latency_ms < 500"

!!! warning SLOs must be explicitly approved (approved: true) before Murat generates test stories.

See SLO Contract for the full contract format.


Winston (Architect)

Role: Consulted on collector architecture decisions, pipeline design, system-wide strategy.

When O11y Engineer Consults Winston

Decision Context
Collector deployment topology Agent vs gateway vs sidecar
Pipeline design Multi-tenant, fan-out configurations
Resource budgets Cluster resources for observability
Technology selection Collector components with broad impact

John (Product Manager)

Role: Owns PRD creation for observability changes, manages change management workflow.

Change Management Flow

  1. O11y Engineer identifies need for change
  2. John creates/approves PRD at _bmad-output/prd/
  3. O11y Engineer generates companion epic
  4. Bob plans the sprint
  5. Amelia implements

!!! note The O11y Engineer will never implement observability changes without a PRD.


Mary (Business Analyst)

Role: Provides requirements for observability use cases, ensures stakeholder alignment.

Collaboration

Mary helps define the use cases section of observability specs:

use_cases:
  - question: "What is the p95 latency for checkout?"
    signal: traces
    answer_method: "Filter spans by service + operation, compute p95 duration"

Mary validates that these questions align with business stakeholder expectations.


Interaction Flow

sequenceDiagram
    participant Mary as Mary (BA)
    participant John as John (PM)
    participant O11y as O11y Engineer
    participant Winston as Winston (Arch)
    participant Bob as Bob (SM)
    participant Amelia as Amelia (Dev)
    participant Murat as Murat (Test)

    Mary->>O11y: Use case requirements
    O11y->>Winston: Architecture consultation
    Winston-->>O11y: Collector topology decision
    O11y->>O11y: Generate observability spec
    O11y->>John: PRD for change management
    John-->>O11y: PRD approved
    O11y->>Bob: Companion epic + stories
    O11y->>Murat: SLO contract
    Bob->>Amelia: Sprint with o11y stories
    Amelia->>Amelia: Implement with spec_reference
    Murat->>Murat: Design tests from SLO contract
    O11y->>O11y: Validate traces against spec
    O11y->>Bob: Fix stories (if failures found)
    Bob->>Amelia: Fix stories in next sprint