Prompt for Comparing AI Coding Agent Efficiency

You are an AI coding agent benchmarking specialist. I need you to design and execute a systematic comparison of AI coding agents. ## Task Create a structured evaluation framework for comparing AI coding agents across these dimensions: ### Agents to Compare - Claude Code - Cursor - Maki - OpenAI Codex CLI - Aider - [Add any others relevant] ### Evaluation Dimensions 1. **Token Efficiency**: Context window usage per task, tokens consumed per successful code change 2. **Speed**: Time-to-first-token, total completion time, startup latency 3. **Code Quality**: Correctness rate, test pass rate, code style adherence 4. **Tool Use**: File navigation strategy, search efficiency, edit precision 5. **Cost**: Estimated cost per task at standard API pricing ### Test Tasks (design 5 representative tasks) 1. Bug fix in a 500-line Python file 2. Add a new API endpoint with tests 3. Refactor a class hierarchy (3+ files) 4. Write documentation from code 5. Debug a failing CI pipeline ### Output Format For each agent, produce: - Quantitative scores (1-10) per dimension - Token usage breakdown (input/output/total) - Strengths and weaknesses summary - Best-fit use case recommendation - Overall ranking with justification Present results as a comparison table followed by detailed analysis per agent. Include methodology notes so the evaluation is reproducible.