David Baker Effendi

Building SlopCop, incremental static analysis for every code change

DavidBaker10005.jpg

I build SlopCop, an online static analysis tool that runs incremental, diff-based checks. Rather than re-reviewing a whole repository, SlopCop checks each change against the commit it builds on, so feedback arrives while the change is still being written, whether from the command line or inside a coding agent.

Static checks surface leads; deeper AI review examines the surrounding code and proposes fixes. Findings keep their source evidence and coverage limits visible, so developers and coding agents can check the diagnosis, fix the problem, or set aside a false positive.

Underneath SlopCop is Bifrost, the open-source, multi-language static-analysis toolbox I also develop. It gives coding agents and developer tools a structural view of unbuilt or partially broken repositories, including mixed-language workspaces. The engine is available as a Rust crate, MCP server, LSP, CLI, and Python client, with queries for code structure and relationships such as usages, imports, and type hierarchies.

How I got here

I came to coding agents through program analysis. My PhD at Stellenbosch University explored how language-agnostic representations, graph backends, and parallel processing could make static analysis more practical at repository scale. Along the way, I maintained Plume and contributed to Joern, working directly with the opportunities and limitations of code property graphs.

From 2022 to June 2025, I applied that research at Whirly Labs, leading the development of tailored static-analysis systems for clients. That work shifted my focus from analysis as a research artifact toward code intelligence as dependable product infrastructure: tools that must remain useful on large, evolving, and occasionally broken codebases.

Since 2025, I have been bringing those lessons to AI-assisted software engineering at SlopCop. SlopCop and Bifrost connect my work in static analysis and program verification to practical code review: finding a suspicious pattern in a change, checking the evidence, and helping people and coding agents make a useful fix.

I remain connected to Stellenbosch University as a visiting lecturer and postgraduate co-supervisor in static program analysis.