| name | report |
| description | This skill should be used when the user invokes /perfmind:report, says "generate a performance report", "write a summary for my manager", "create a performance write-up for leadership", "show me recommendations as a DevOps engineer", "give me actionable fixes as a developer", "write an analyst-level summary", "I need to share these findings with the team", "write a report for my team", "document these perf findings", "share findings with stakeholders", or asks to document performance findings for a specific audience. Reads investigation findings from the conversation and auto-detects the target audience. |
| argument-hint | [role] e.g. developer | devops | leadership | perf-engineer |
| allowed-tools | Read |
Performance Report
Role-Tailored Performance Report Generation
Generate a structured performance report adapted to the target audience.
Step 1: Gather Findings
Read the performance findings from the current conversation context. If no /perfmind:investigate session has been run, prompt the user to share key findings first or invoke /perfmind:investigate before proceeding.
Step 2: Detect Target Role
If the user passes a role argument, use it directly. Otherwise, infer from conversation context:
| Signal words | Role |
|---|
| "deploy", "infra", "Kubernetes", "scaling", "SRE", "runbook", "alert" | DevOps |
| "code fix", "PR", "refactor", "function", "file", "commit" | Developer |
| "flame graph", "profiler", "benchmark", "regression", "p99", "percentile" | Perf Engineer |
| "SLA", "user experience", "cost", "roadmap", "business impact", "budget" | Leadership |
If unclear, default to Developer.
Step 3: Generate Role-Specific Report
Generate the report using the full template for the detected role from references/role-templates.md.
Key output principles per role:
| Role | Focus | Tone | Length |
|---|
| Developer | Code-level fixes, file:line references, patch-sized items | Technical, specific | No limit |
| Perf Engineer | Baseline metrics, evidence chain, data gaps, regression prevention | Analytical, metric-first | No limit |
| DevOps | SLO status, infrastructure actions, runbook entry | Operational, config-focused | No limit |
| Leadership | Business impact, effort vs. timeline, risk if not addressed | Non-technical, outcome-focused | <400 words |
After generating the report, output: "Share more evidence via /perfmind:investigate to refine these recommendations."
Additional Resources
references/role-templates.md — Full markdown templates for all four roles with section-by-section guidance and audience notes