Deprecated: Agent Enforcer Open Beta: Enforce the Quality of Agent Code
A reference architecture for building robust, language-agnostic local CI systems.
Agent Enforcer 2 is not a tool you install — it’s a concept and blueprint for implementing local CI in your projects. It documents battle-tested patterns for orchestrating code quality checks, managing caches, detecting hangs, and producing structured reports.
Last summer I tried to publish my approach to building a more automated dev workflow than just copy-pasting errors into an LLM
When I get error messages I just copy paste them in with no comment
I decided to make the system more LLM-ish than how I personally use it, by adding the ability to call a tool via MCP; but the then-current Gemini 2.5 Pro refused to run when MCP tools were connected, same as some other models. I also wanted to make a universal thing for different languages, but luckily I didn’t have enough time for that back then. And I’m still creating my own local CI for each project, even though I base it on the CI foundation from my latest projects.
Now I decided to make it easier for myself and focus on describing the concept I follow. Because if we’re talking about AI-assisted development, agents generate code fast and can generate it tailored to the target project — we just need to provide the rules.
AE2 is how I currently approach the concept of building local CI in my projects. I’d even be interested in running a comparative test of whether LLMs can achieve better results on something like SWE-Bench Pro with AE2 and Agent Compass, but I don’t have the resources to run a comprehensive test ¯\_(ツ)_/¯