The tool, named after the mariner's astrolabe, is highly specialised to Cloudflare's internal systems and is not available to customers. It performs scans in two stages: a discovery stage that maps the repository and searches for cryptography across source, configuration, manifests, lockfiles, scripts, tests and documentation, and an analysis stage that enriches findings by pulling information from internal documentation and ticketing systems. The company says that traditional methods like grepping for algorithm names overcount or undercount, and cannot explain how the cryptography is used, whereas AI can follow evidence across files and return structured analysis.

Cloudflare faces three main challenges in its migration: its code is spread across many repositories, cryptography rarely announces itself plainly in the code, and discovery is about more than just pattern matching. The company also wants to provide progress metrics for the migration and surface prerequisites early, such as protocols that don't yet have a PQ migration plan. This helps it work with standards bodies and ecosystems to meet its 2029 timeline.

The tool is still evolving as development continues, but Cloudflare is sharing its learnings so other organisations can build upon its efforts as they work through their own PQ migration journeys.