Limitations

Genuine limitations

• ForkReason analyzes public GitHub repositories only in V1.
• A repository with a squashed or shallow history has weak chronology, and the report will say the history was limited.
• A monorepo is analyzed as one repository; it is not decomposed into independently versioned packages.
• Shared-upstream discovery is heuristic for repositories that do not declare a fork parent, and can miss a common ancestor.
• Authorship, ownership and intent are never inferred. ForkReason reports development relationships, not who wrote what or why.
• Gencode, minified output and vendored trees are inventoried but not analyzed, because they carry no authorship signal.

Deliberate non-goals

• No legal conclusions. ForkReason never says a repository was stolen or infringing.
• No ownership verification.
• No blame. A finding describes a relationship, not a person.
• No confidence percentages, because there is no calibration methodology.

A tool that claims to measure provenance honestly has to state where it is not measuring it.