ResourceGraph, from orphan fields to trust boundaries
- hypothesisevidence level
- decision
- 2026-11-15
- #vulnmind
- #static-analysis
- #openapi
ResourceGraph parses OpenAPI specifications into a graph of resources and looks at how ownership is expressed on each one.
The signal
If a resource has an ownership field (say, which account it belongs to) but the spec does not make that field required, then some code path may be able to create or reach the resource without an owner being enforced. That is a possible authorization weakness, so ResourceGraph flags it.
How the framing changed
The first version was a cruder heuristic: flag any “orphan” field with no obvious relationship. It produced noise because most orphan fields are harmless. Reframing the question as a trust-boundary question, namely whether the spec requires ownership where it exists, made the output far more specific.
Status of the claim
This is a hypothesis. A flag means “worth a look”, not “vulnerable”. No detection rate, false-positive rate or real-world hit count is claimed here. Those need to be measured before this document can move up an evidence level.