AeonScript is a layered, self-describing format for writing information onto DNA — engineered to survive millennia, readable by any conformant decoder, owned by no one.
We produce 120 zettabytes of data per year. We can store ~10 % of it. Of what we store, almost nothing will be readable a century from now. The medium itself decays — and the formats become orphaned.
Today, every major DNA storage vendor — Microsoft, Twist, Catalog — uses a different proprietary encoding. A file written by one cannot be read by another. AeonScript is the open standard that fixes this, before the market is captured.
Like OSI for networking. Each layer can evolve without breaking the others. The wire format is self-describing — a 2030 reader will still parse a 2230 file, because the file itself declares its codec.
implemented in v0.1 reference codec · planned for v0.2
Type a message — or drop an image. AeonScript encodes bytes, not text, so any file works the same way. The L1 + L5 encoder runs locally in your browser — no server, no data leaves this page.
Click "Encode" to see the oligos.
Curious about that |AEONSCRIPT=0.1;...| tag at the top? Dissect it field by field →
Reference implementation: Python · Node.js · Same algorithms as below.
# Install $ pip install -e git+https://github.com/aeonscript-spec/aeonscript@main#subdirectory=reference # Use >>> from aeonscript import encode_file, decode_oligos >>> oligos = encode_file("my_archive.txt") >>> decode_oligos(oligos) == open("my_archive.txt", "rb").read() True
Spec under CC-BY-SA, code under MIT. No royalties, no NDAs, no premium tier.
Each block carries a semantic tag declaring its own codec version. No manual required in 2230.
Reed-Solomon inside, LDPC across blocks, methylation-channel hash for silent-corruption detection.
Conformant encoders must screen against pathogen databases (IGSC) before emitting any oligo for synthesis.
The L4+ codec is designed to be driven by genomic foundation models (Evo 2, Caduceus) for near-Shannon density.
Designed for adoption by libraries, archives, and standards bodies — not vendors.
v0.1 is an early draft. The L1 / L4 / L5 layers are normatively specified and have a working reference implementation. L2 / L3 / L6 / L7 are described at the interface level only. See the roadmap for what's next.
If you work in bioinformatics, theory of information, archival science, or you simply believe the memory of humanity shouldn't be proprietary — there is a place for you here.