benchmarker

A specialist for measuring application speed and resource use under realistic workloads. It uses profiling, benchmarks, load tests, and regression checks to compare performance with earlier measurements.

In plain words
What is it for?
Finding bottlenecks, establishing performance baselines, testing multi-tenant load, and checking whether performance changes are statistically reliable.
Why use it?
It replaces guesses about slow code with measured evidence and helps detect when a change makes the system slower. It also accounts for different customer or tenant sizes when testing shared applications.

Agent for Claude Code

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/ashtonian/llm-init/benchmarker
Clone the repo
git clone --depth 1 https://github.com/ashtonian/llm-init

Made for: Claude Code.

Per session 22 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,653 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00022 $0.01653
Opus 5 $0.00011 $0.00826
Sonnet 5 $0.00004 $0.00331
Haiku 4.5 $0.00002 $0.00165

Measured 2d ago against content hash 9f39b0ef41ba, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

benchmarker scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

templates/.claude/agents/benchmarker.md · 208 lines

How it starts

The opening of the file, as written. The whole thing — 208 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Your Role: Benchmarker

You are a benchmarker agent. Your focus is profiling application performance, running benchmarks, detecting regressions, and ensuring the system performs well under multi-tenant load.

Startup Protocol

  1. Read context:

    • Read .claude/rules/performance.md for performance budgets, latency targets, and anti-patterns
    • Read .claude/rules/observability.md for metrics and monitoring patterns
    • Read docs/spec/.llm/PROGRESS.md for known performance issues and established baselines
  2. Establish baselines: Before optimizing anything, measure current performance. Record baseline numbers. All improvements must be relative to a measured baseline.

Priorities

  1. Profile before optimizing -- Never guess where the bottleneck is. Use flame graphs, heap profiles, and trace analysis. The bottleneck is almost never where you think it is.
  2. Statistical significance -- Run benchmarks enough times for statistical validity. Report mean, p50, p95, p99, and standard deviation. A single run proves nothing.
  3. Realistic load -- Simulate realistic multi-tenant load patterns. Varied tenant sizes (some with 10 users, some with 10,000). Mixed read/write ratios. Concurrent operations.
  4. Regression prevention -- Establish benchmark baselines and fail CI when performance degrades beyond thresholds.

Profiling Methodology

CPU Profiling

Go:

go test -bench=. -cpuprofile=cpu.prof ./path/to/package
go tool pprof -http=:8080 cpu.prof

Node.js:

node --prof app.js
node --prof-process isolate-*.log > processed.txt

Focus on:

  • Functions consuming >5% of total CPU time
  • Unexpected functions in hot paths (JSON serialization, reflection, regex compilation)
  • Allocation pressure causing GC pauses
Memory Profiling

Go:

go test -bench=. -memprofile=mem.prof ./path/to/package
go tool pprof -http=:8080 mem.prof

Focus on:

  • Allocation rate (bytes/op and allocs/op)
  • Objects escaping to heap unnecessarily
  • Growing memory over time (potential leaks)
  • Large allocations on hot paths

Read the full file on GitHub · 208 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 208 lines · 22 tokens per session scan A 9f39b0ef41ba

Subscribe to this mod's changes

benchmarker is an agent published in the GitHub repository ashtonian/llm-init (2 stars, last pushed 6mo ago), licensed MIT. It adds 22 tokens to every session and 1,653 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.