Multi Agent Performance Profiling
Analyzes and optimizes collaborative agent teams' performance via detailed metrics and actionable insights from Terrylica.
Install on your platform
We auto-selected Claude Code based on this skill’s supported platforms.
Run in terminal (recommended)
claude mcp add multi-agent-performance-profiling npx -- -y @trustedskills/multi-agent-performance-profiling
Or manually add to ~/.claude/settings.json
{
"mcpServers": {
"multi-agent-performance-profiling": {
"command": "npx",
"args": [
"-y",
"@trustedskills/multi-agent-performance-profiling"
]
}
}
}Requires Claude Code (claude CLI). Run claude --version to verify your install.
About This Skill
What it does
This skill analyzes and optimizes the performance of collaborative agent teams within complex workflows. It uses a multi-layer profiling approach involving five parallel agents to identify bottlenecks across various system layers, such as database configuration, client library usage, batch sizes, and overall pipeline stages. The skill quantifies each stage's contribution to total time and prioritizes optimization efforts based on impact.
When to use it
- When performance is below a defined Service Level Objective (SLO).
- For optimizing multi-stage pipelines (e.g., download → extract → parse → ingest).
- To investigate database performance issues.
- To identify bottlenecks in complex workflows.
- As a pre-optimization analysis step before making changes to a system.
Key capabilities
- Multi-Layer Profiling: Uses five parallel agents for comprehensive performance assessment.
- Phase-Boundary Instrumentation: Measures time using
time.perf_counter()at key stages in the workflow. - Memory Profiling: Tracks peak memory usage and allocations.
- Database Configuration Analysis: Reviews server settings and compares production vs. development configurations.
- Client Library Analysis: Examines API usage patterns and identifies buffer size tuning opportunities.
- Batch Size Analysis: Determines optimal batch sizes considering memory overhead and throughput.
- Prioritization: Assigns priorities (P0/P1/P2) to optimization tasks based on impact quantification.
Example prompts
- "Analyze the performance of our data ingestion pipeline."
- "Identify bottlenecks in our multi-stage processing workflow."
- "Profile the database configuration and suggest potential improvements."
Tips & gotchas
- The skill operates using a parallel execution pattern, with no dependencies between investigation agents (1-4). The integration agent synthesizes findings from all agents.
- It's designed to identify true bottlenecks rather than relying on assumed causes.
- Dynamic todo management is used; the skill spawns agents in parallel and updates todos as they complete.
Tags
TrustedSkills Verification
Unlike other registries that point to live repositories, TrustedSkills pins every skill to a verified commit hash. This protects you from malicious updates — what you install today is exactly what was reviewed and verified.
Security Audits
| Gen Agent Trust Hub | Pass |
| Socket | Pass |
| Snyk | Pass |
🌐 Community
Passed automated security scans.