agent-sh

Perf Theory Tester — Development agent for Claude Code

Development community

Execute controlled perf experiments, one change at a time, with rollback between runs.

How to install Perf Theory Tester

Installs to ~/.claude/agents/agent-sh-perf-perf-theory-tester.md

Terminal
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/agent-sh/perf/HEAD/agents/perf-theory-tester.md -o ~/.claude/agents/agent-sh-perf-perf-theory-tester.md

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

What Perf Theory Tester does

name: perf-theory-tester description: Execute controlled perf experiments, one change at a time, with rollback between runs. tools:

  • Skill
  • Read
  • Write
  • Edit
  • Bash(git:*)
  • Bash(npm:*)
  • Bash(pnpm:*)
  • Bash(yarn:*)
  • Bash(cargo:*)
  • Bash(go:*)
  • Bash(pytest:*)
  • Bash(python:*)
  • Bash(mvn:*)
  • Bash(gradle:*)
  • Bash(node:*) model: opus

Perf Theory Tester

Test hypotheses using controlled experiments. You MUST follow `docs/perf-requirements.md`.

You MU

Alternatives in Development

  • Eval Engineer — GAIA evaluation framework specialist 1.5k ★
  • Leader — Central decision-maker that plans experiments and reflects on results 728 ★
  • Theorist — Theoretical econometrician / mathematical statistician 250 ★

Full documentation available on GitHub

View Source Repository

Related Agents

Selfish Evolution — Development agent for Claude Code

Selfish Evolution

Run retrospectives and controlled experiments that improve the system while preserving the human’s authority o

Development community
Noemeo Researcher — Development agent for Claude Code

Noemeo Researcher

Plans, runs and writes up studies in Noemeo repos -- falsifiable questions, prior-work checks, public benchmar

Development community
Perf Theory Gatherer — Development agent for Claude Code

Perf Theory Gatherer

Generate top performance hypotheses after reviewing git history and current metrics.

Development community
Prototyper — Development agent for Claude Code

Prototyper

The Prototyper builds rapid proof-of-concept implementations, throwaway spikes, and time-boxed experiments to

Development community
Promotion Officer — Development agent for Claude Code

Promotion Officer

Use when authorizing transition between subscription modes (start_shadow shadow_to_assisted assisted_to_autono

Development community
Perf Tester — Development agent for Claude Code

Perf Tester

Use to measure latency, throughput, and load behaviour of HTTP endpoints or hot functions with autocannon or k

Development community

Related Skills

Dev Studio — Testing skill for Claude Code

Dev Studio

Studio v2 umbrella router. Dispatches by canonical role from any project: manager, worker, reviewer, perf, pla

Testing community
Execute Plan Attended — Productivity skill for Claude Code

Execute Plan Attended

Execute a plan one phase at a time; present evidence and wait for explicit approval between phases

Productivity community
Perf Run — Testing skill for Claude Code

Perf Run

Execute a perf test type (load, stress, spike, soak, breakpoint) with pre-flight checks, then analyze results.

Testing community