| schemaVersion | "2026-04-11T00:00:00.000Z" |
| skillId | backend/backend-performance-profiling |
| name | backend-performance-profiling |
| displayName | Backend Performance Profiling |
| description | Use when working on backend performance profiling, slow endpoint analysis, query hotspots, CPU, memory, and latency work. Focus on measurement, profiling evidence, hot path reduction, and regression guards. |
| aliases | ["backend-performance-profiling","Backend Performance Profiling","backendperformanceprofiling","服务端","server side","性能","性能优化","性能分析","profiling","调试","排查","定位问题"] |
| version | 0.1.0 |
| sourceHash | sha256:c0350215c0c4f70e4896b2095f7faa56415ccafd71adf131733f6a077a9cbef9 |
| domain | backend |
| departmentTags | ["backend-platform"] |
| sceneTags | ["debug","performance"] |
Backend Performance Profiling
Use this skill when the task involves backend performance profiling, slow endpoint analysis, query hotspots, CPU, memory, and latency work.
Goal: produce reliable engineering guidance and implementation steps focused on measurement, profiling evidence, hot path reduction, and regression guards.
Working model
- Identify the affected system, data, users, and failure modes.
- Define invariants, inputs, outputs, ownership, and rollback needs.
- Prefer small, auditable changes with explicit validation.
- Call out security, performance, concurrency, and data-loss risks when relevant.
- Finish with concrete verification steps and residual risks.
Rules
- Ground recommendations in the current codebase or runtime evidence.
- Prefer explicit contracts, typed boundaries, and defensive validation.
- Do not hide operational concerns behind generic best practices.
- Include negative cases, edge cases, and failure behavior.
- For review tasks, list findings first with file and line references when possible.
- For test or performance tasks, define the workload, success criteria, and measurement method.
Checklist
- Are assumptions and ownership boundaries explicit?
- Are risky changes reversible or safely deployable?
- Are observability and diagnostics sufficient for production issues?
- Are tests or validation steps targeted to the actual risk?
- Are security and data-integrity concerns addressed?
