The O11y Engineer follows an 8-phase workflow to take a project from zero to production-grade observability. Each phase builds on the previous one.
graph LR
P1[1. Assessment] --> P2[2. Observability Spec]
P2 --> P3[3. Collector Config]
P3 --> P4[4. CI/CD Weaver]
P4 --> P5[5. SLO Definition]
P5 --> P6[6. MCP Rules]
P6 --> P7[7. Implementation]
P7 --> P8[8. Validation]
P8 -->|Fix stories| P7
style P1 fill:#2563eb,color:#fff
style P2 fill:#7c3aed,color:#fff
style P3 fill:#059669,color:#fff
style P4 fill:#d97706,color:#fff
style P5 fill:#dc2626,color:#fff
style P6 fill:#0891b2,color:#fff
style P7 fill:#be185d,color:#fff
style P8 fill:#4f46e5,color:#fff
| Phase | Name | Command | Output |
|---|---|---|---|
| 1 | Assessment | AM via /o11y-engineer |
Maturity score, gap analysis, roadmap |
| 2 | Observability Spec | Design spec with agent | Spec file + companion epic |
| 3 | Collector Config | CP via /o11y-engineer |
Collector YAML, processor ordering |
| 4 | CI/CD with Weaver | VS via /o11y-engineer |
Weaver config, CI pipeline |
| 5 | SLO Definition | Define SLOs with agent | SLO contract, test stories |
| 6 | MCP Rules | Configure MCP rules | IDE rule files with DQL |
| 7 | Implementation | /o11y-generate-epics |
Stories assigned to dev agents |
| 8 | Validation | VO via /o11y-engineer |
Validation report, fix stories |
Goal: Understand the current state of observability in the project.
The assessment scans existing instrumentation, collector configuration, and monitoring setup to produce a maturity score (0-100) and a prioritized improvement roadmap.
Key questions answered:
Goal: Define what telemetry each service should produce and why.
The spec is a use-case-driven document that serves as a contract between development and operations. It defines trace contracts, log contracts, metric contracts, and correlation contracts.
Key outputs:
observability-specs/{service}-spec.yamlGoal: Design and configure the OpenTelemetry Collector pipeline.
This phase covers receiver selection, processor ordering (memory_limiter first, batch last), resource enrichment, PII cleanup, and exporter configuration.
Key decisions:
Goal: Integrate OpenTelemetry Weaver into the CI/CD pipeline for automated semantic convention validation.
Weaver ensures that any telemetry changes are validated against the defined semantic conventions before they reach production.
Goal: Define performance, reliability, and availability targets based on the observability spec.
SLOs are presented for user approval before generating test stories. Each SLO maps to a specific test type (load, soak, chaos, synthetic) for the Test Architect.
Goal: Generate IDE-specific rule files that embed Dynatrace DQL queries.
Rule files are placed in the project directory according to IDE conventions (Claude Code, Cursor, Windsurf, Copilot, Amazon Q).
Goal: Hand off to the B-MAD agent ecosystem for implementation.
The O11y Engineer generates stories in BMAD standard format. Bob (Scrum Master) plans sprints, Amelia (Dev) implements instrumentation, and Murat (Test Architect) designs tests from the SLO contract.
Goal: Verify that actual telemetry matches the observability spec.
The validation workflow runs the mandatory 5-step Query Validation Gate before concluding any span/metric/log is missing. It produces dual-format reports (human-readable markdown + machine-readable YAML) and generates fix stories for failures.
| Scenario | Start At |
|---|---|
| Greenfield project, no observability | Phase 1 |
| Existing observability, needs improvement | Phase 1 |
| Collector already configured | Phase 2 |
| Specs already defined | Phase 3 |
| Ready for SLOs | Phase 5 |
| Need to validate existing setup | Phase 8 |
The workflow is iterative. After Phase 8 validation:
graph TD
V[Phase 8: Validation] --> D{All Pass?}
D -->|Yes| Done[Production Ready]
D -->|No| Fix[Generate Fix Stories]
Fix --> P7[Phase 7: Implementation]
P7 --> V