performance-benchmarker — Claude Code subagent
performance-benchmarker is a Claude Code subagent that measures throughput, latency, and resource use to compare Byzantine, Raft, and Gossip consensus protocols.
no license ★ 850 updated 2026-10-26
.claude/agents/performance-benchmarker.mdWhat performance-benchmarker does
The subagent provides a complete benchmarking toolkit: generates protocol load, tracks transaction throughput and latency, monitors CPU, memory, and network usage. It stores results in a time-series database for historical tracking and produces comparative analysis across protocols. An adaptive optimizer adjusts load parameters in real time based on test success rates.
Useful when building and tuning distributed systems: the subagent automatically executes full benchmark cycles on hook triggers, measures latency percentiles (p95, p99), and analyzes fault tolerance. Results are timestamped so you can detect performance degradation or improvements after code changes.
What it helps with
- measure throughput of consensus algorithms under varying load
- compare latency across Byzantine, Raft, and Gossip protocols
- track resource usage (CPU, memory, network) during distributed consensus
- identify bottlenecks and optimize protocol parameters automatically
- maintain a searchable history of benchmark runs
Example prompts for the agent
- Run performance-benchmarker to compare Byzantine and Raft at 1000 transactions per second
- Collect latency and throughput metrics for Gossip protocol while varying node count
- Analyze Raft fault tolerance when one validator fails and compare against Byzantine
How to install performance-benchmarker
Manually: put the files in place
.claude/agents/performance-benchmarker.mdFor all projects use ~/.claude/agents/. Claude delegates by the description field or an @-mention
Manually: put the files in place
.codex/agents/performance-benchmarker.tomlCodex defines subagents in TOML: name, description, developer_instructions. Move the Markdown text into developer_instructions by hand. For all projects use ~/.codex/agents/
Manually: put the files in place
.cursor/agents/performance-benchmarker.mdCursor also reads .claude/agents/
Step by step, with commands for each platform: Claude Code subagents: how to create one, call it and keep context clean
Questions about performance-benchmarker
How do I install performance-benchmarker in Claude Code?
Add the file .claude/agents/performance-benchmarker.md to your project root. Claude Code will automatically load the subagent and make it available for benchmarking tasks.
What does performance-benchmarker do?
It measures throughput, latency, and resource consumption for distributed consensus protocols. Generates synthetic load, collects metrics, compares results across protocols, and recommends optimizations.
Which consensus protocols does it support?
Byzantine, Raft, and Gossip. The subagent lets you register new protocols by adding benchmark suites.