Coding agents finish tasks and pass tests, yet repos get harder to change after many iterations. CXCAP is a free, open-source CLI that shows what a change will touch before anyone edits: likely files, dependents, import cycles, cross-boundary edges, and where static analysis can't see. Local, read-only, no model or account. Python, JS/TS, Rust. Installs an agent skill so Claude Code and other agents can call it. Try to break it: run it on your most AI-built repo.

I built CXCAP because I was using frontier AI coding agents heavily and kept seeing the same failure mode. Individual tasks succeeded. Tests passed. Each iteration looked reasonable. But after enough iterations, the codebase became progressively harder to work with. The AI agents were strong at the task directly in front of them, but they had no independent structural feedback loop telling them what complexity already surrounded the next change. That also started to matter economically: the harder the repository became to understand, the more context, reasoning, and AI agent work future changes required. I wanted something outside the model—local, deterministic, and read-only—that could inspect the repository before another change was made. That became CXCAP.

CXCAP

The complexity feedback loop for AI coding agents

The complexity feedback loop for AI coding agents