llm-cost-analysis

Analyze ai_call_logs to identify costly LLM calls and latency bottlenecks.

19|4|Updated Jan 11, 2026

Search Tags:#logs #optimization #llm #latency #analyze #cost #ai-calls

Ready to run

npx skills add https://github.com/ZanebonoAlter/Syntopica --skill llm-cost-analysis

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: llm-cost-analysis
Source: https://github.com/ZanebonoAlter/Syntopica/tree/main/.claude/skills/llm-cost-analysis
Command: npx skills add https://github.com/ZanebonoAlter/Syntopica --skill llm-cost-analysis

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Analyze LLM call frequency, latency, and success rate across capabilities to identify optimization opportunities.

Core Features & Use Cases

Global overview of LLM usage across capabilities with key metrics (calls, average latency, success rate)
Per-operation hotspots: identify top operations (e.g., topic_tagging) that drive cost or latency
Actionable guidance: map findings to optimization steps and best practices

Quick Start

Run an analysis over ai_call_logs to identify the top costliest and slowest operations, then review recommended optimizations.

Frequently Asked Questions about llm-cost-analysis

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I analyze LLM call logs to identify cost and latency bottlenecks?▼

To analyze LLM call logs, evaluate ai_call_logs to identify costly operations, latency hotspots, and success-rate patterns across capabilities like topic_tagging, tagging-flow, and embedding.

What is the best way to diagnose high latency in AI calls?▼

The best way to diagnose AI call latency is to analyze per-operation hotspots in your ai_call_logs, isolating specific workflows like topic_tagging that drive delays, then reviewing recommended mitigations.

How do I find which LLM operations are driving up API costs?▼

You can find costly LLM operations by running a global overview of your LLM usage to rank per-operation hotspots, exposing which specific capabilities consume the most resources and drive costs.

Can I identify error patterns and success rates across different AI capabilities?▼

Yes, you can identify error patterns by analyzing success-rate metrics across capabilities within your ai_call_logs, mapping failure trends to actionable optimization steps and best practices.

Does this LLM usage analysis require specific log formats to work?▼

This LLM usage analysis specifically targets ai_call_logs containing call frequency, latency, and success rate data to accurately locate operational hotspots and provide optimization guidance.