Experiment — Development agent for Claude Code
Performs focused technical experiments for uncertain behavior, algorithms, geometry, rendering, or performance.
How to install Experiment
Installs to ~/.claude/agents/debasishrana14-claude-agents-experiment.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Debasishrana14/Claude-Agents/HEAD/.claude/agents/experiment.md -o ~/.claude/agents/debasishrana14-claude-agents-experiment.md Restart Claude Code, or start a new session, for it to be picked up.
What Experiment does
name: experiment description: Performs focused technical experiments for uncertain behavior, algorithms, geometry, rendering, or performance. Use only when static inspection cannot confidently answer the question. model: sonnet
Experiment Agent
Role
Answer one technical uncertainty using a controlled experiment.
Before Starting
Define:
Hypothesis: Input: Expected result:
Preferred Methods
Use:
- temporary scripts
- isolated tests
- minimal reproductions
- be
Alternatives in Development
- Eval Engineer — GAIA evaluation framework specialist 1.5k ★
- Leader — Central decision-maker that plans experiments and reflects on results 728 ★
- Geometry Change — Use when modifying the keyboard rendering pipeline — the files src/lib/worker/geometry.ts, src/lib/worker/cach 662 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
3D Graphics Specialist
Use this agent when you need expertise in 3D rendering, geometry algorithms, WebGL optimization, or shader pro
Implementer Hard
Implements hard code changes end to end (Opus, high effort). Use for scoped but difficult work that can run wi
Autoresearch Orchestrator
Use this agent to run an autoresearch experiment session end-to-end — setup, baseline, and a batch of edit→mea
Evaluation Orchestrator
Use for coordinating evaluation experiments, managing experiment execution, and overseeing evaluation workflow
Trainer
Runs training experiments for tiny models — baseline first, then the model, respecting size/latency/wallclock
LinkedIn Performance Review
Analyzes LinkedIn post results, maintains experiments, and proposes evidence-based learning (/review). Use whe
Related Skills
Agent Explain Codebase
A focused brief that makes a coding agent map an unfamiliar codebase so you (or it) can work in it confidently
Android Layout Inspector MCP
Spatial inspection of a live Android UI for AI agents — finds overlapping views, clipping, touch targets under
Generating Business Collateral
Claude Code skill for producing a business's onsite print collateral and online assets as verified files — der