jgreathouse9

Agents Benchmarking — Development agent for Claude Code

Development community

agents_benchmarking.md.

How to install Agents Benchmarking

Installs to ~/.claude/agents/jgreathouse9-mlsynth-agents_benchmarking.md

Terminal

mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/jgreathouse9/mlsynth/HEAD/agents/agents_benchmarking.md -o ~/.claude/agents/jgreathouse9-mlsynth-agents_benchmarking.md

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

What Agents Benchmarking does

agents_benchmarking.md

Operational guide for adding **durable, re-runnable** validation to an mlsynth estimator. This is the *process* doc (the "why" and the definitions of done); `benchmarks/README.md` is the *mechanics* doc (the "what to run"), and `docs/replications.rst` is the public catalogue. Read `.claude/commands/`'s `/replicate` for the demonstrate-first loop that *precedes* the work described here.

A benchmark is not finished when it prints a number that looks right once. It is fin

Alternatives in Development

Eval Engineer — GAIA evaluation framework specialist 1.5k ★
Bench — API performance benchmarking — latency profiling, throughput testing, performance regression detection 73 ★
Swe Perf Reviewer — Performance reviewer that identifies computational bottlenecks, benchmarking gaps, and optimization opportunit 18 ★