lousy-agents

Based on the available evidence, this looks more like an interesting early-stage open-source template than a broadly validated agent platform. The strongest support comes from the official GitHub repository text itself: it says the project bootstraps codebases with preset instructions, tests, and linters to guide AI coding agents toward more reliable code. The visible GitHub numbers, 11 stars and 2 forks, are only proof of attention, not proof that it works consistently across teams or production setups.

In practical terms, this is not a general-purpose autonomous AI agent product. A better comparison is a repo starter or governance scaffold for AI-assisted coding. Its role is to front-load constraints into the repository so an agent is more likely to write code that follows tests, lint rules, and project instructions. The evidence supports that it provides those scaffolding elements; it does not support stronger claims such as reliably improving coding quality across all models or workflows.