klh

LLM Performance Analyst — Data & AI agent for Claude Code

Data & AI community

Evaluates LLM agent performance from session transcripts — token usage, tool-call efficiency, and error patterns.

How to install LLM Performance Analyst

Installs to ~/.claude/agents/klh-speedy-claude-llm-performance-analyst.md

Terminal

mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/klh/speedy-claude/HEAD/agents/llm-performance-analyst.md -o ~/.claude/agents/klh-speedy-claude-llm-performance-analyst.md

Restart Claude Code, or start a new session, for it to be picked up.

What LLM Performance Analyst does

name: llm-performance-analyst description: Evaluates LLM agent performance from session transcripts — token usage, tool-call efficiency, and error patterns. Diagnoses dumb errors from hasty narrow edits that miss the bigger pattern, and prescribes concrete fixes to CLAUDE.md, hooks, permissions, and skills. Scope is agent behavior and agent config only — auditing online services or SaaS subscriptions is tool-stack-auditor's job. Use after long sessions, cost spikes, repeated error loops, or

Alternatives in Data & AI

API Cost Scanner — Scans codebase for AI/API cost optimization opportunities including model usage, token estimates, caching stra 160 ★
Review Code Correctness — A code review agent that checks analysis scripts for logical errors, incorrect function usage, wrong variable 134 ★
Cross Service Auditor — Phase 8 cross-service taint-propagation agent that stitches inter-component data flows (HTTP/gRPC/queues/IPC/s 125 ★