I will benchmark your ai coding agent with a clear test report

AI coding agents can produce convincing output that still misses instructions, breaks existing behavior, or passes only shallow checks.

I will benchmark a bounded agent run against your supplied task and repository, execute relevant tests where feasible, and return an evidence-based report.

You will receive:

scope and test environment

pass, fail, and unknown findings

reproduction steps and test evidence

instruction-following notes

regressions and edge cases

recommendations clearly separated from verified results

Good fit for Codex, Copilot, Claude-style coding agents, autonomous repository tasks, and patch or PR comparisons.

Please provide the repository or relevant files, task or prompt, agent output or patch, available run logs, setup steps, and expected result. Remove API keys, passwords, personal data, and other secrets before sending.

This service is testing and reporting only. It does not include a formal security audit, penetration test, unlimited repository review, production deployment, or a guarantee of model quality.