Context
The current semantic pipeline stores its own function rows and typed similarity evidence. Analysts using existing reverse-engineering tools need an explicit interchange boundary rather than a claim of compatibility with external binary-analysis databases.
Proposal
Define a documented, versioned adapter for importing/exporting supported function boundaries, instruction metadata, CFG summaries, artifact identity, and analysis limitations in a BinExport-compatible or clearly named intermediate format. Start with the subset the corpus can prove and reject unsupported fields explicitly.
Acceptance criteria
- The format documents supported fields, architecture limits, and loss semantics.
- Round-trip tests preserve artifact identity, function offsets, supported CFG metadata, and limitations.
- Imports are tenant-scoped, digest-verified, and cannot create unverified strong variant edges by themselves.
- The CLI reports unsupported or lossy fields instead of silently dropping them.
References
Context
The current semantic pipeline stores its own function rows and typed similarity evidence. Analysts using existing reverse-engineering tools need an explicit interchange boundary rather than a claim of compatibility with external binary-analysis databases.
Proposal
Define a documented, versioned adapter for importing/exporting supported function boundaries, instruction metadata, CFG summaries, artifact identity, and analysis limitations in a BinExport-compatible or clearly named intermediate format. Start with the subset the corpus can prove and reject unsupported fields explicitly.
Acceptance criteria
References