bmad-performance-optimization_skill

Overview

This skill diagnoses runtime bottlenecks and produces prioritized performance optimization plans. I deliver measurable recommendations, load and benchmark plans, and an actionable backlog that balances speed, cost, and complexity. The goal is to keep the product within agreed performance budgets and SLAs while minimizing risk of regressions.

How this skill works

I ingest architecture diagrams, observability data, profiling dumps, and load test reports to locate hotspots across code, database, network, and frontend layers. I validate instrumentation, define acceptance thresholds with stakeholders, and create repeatable benchmark scenarios. Outputs include a performance brief, a ranked optimization backlog, and verification approaches for before/after measurement.

When to use it

• You observe latency, throughput, or resource regressions in production or staging.
• You need a load or performance testing strategy or help interpreting test results.
• You must set or validate performance budgets, SLAs, or SLOs.
• You plan scaling for launches, promotions, or traffic spikes.
• You want guidance on tuning code, queries, caching, or infrastructure for speed.

Best practices

• Always start from reliable telemetry; prioritize fixing instrumentation gaps before major changes.
• Define clear success criteria (metrics and thresholds) for every recommendation.
• Rank fixes by impact versus implementation effort and include an owner and rollout plan.
• Use hypothesis-driven experiments when telemetry conflicts with assumptions.
• Include regression safeguards and automated checks in CI to detect performance drift.

Example use cases

• Analyze a sudden increase in API latency and recommend targeted code or DB fixes with verification metrics.
• Design a load test plan for a product launch covering baseline, stress, soak, and spike scenarios.
• Convert ad-hoc optimization ideas into a prioritized backlog with impact estimates and rollout steps.
• Assess caching and partitioning strategies to meet reduced p99 latency targets while controlling cost.
• Validate proposed architecture changes against projected workload and update SLAs and budgets accordingly.