Auto-Discovering Cross-Repo Infrastructure Dependencies Is Harder Than It Looks

Platform teams often lack visibility into cross-repository infrastructure dependencies, leaving them unable to predict what breaks when a change is made. While the core idea of scanning repos and parsing dependency files to build a directed graph is straightforward, each ecosystem — including Terraform, Docker, CI pipelines, Python, Go, npm, Ansible, Helm, and Kubernetes — introduces its own parsing complexities and edge cases. Many engineers have built homegrown solutions using cron jobs, shell scripts, and lightweight databases, but these tend to go stale, miss edge cases, or collapse when the original author leaves. The pipeline involves enumerating repos, shallow-cloning them, parsing relevant files, detecting artifacts, resolving references to known repos, and storing the result as a queryable graph. The cross-ecosystem challenge is significantly harder than solving any single ecosystem, largely because dependency information is scattered across dozens of file formats with no unified structure.