We study distributed declarative workflow execution in an
adversarial setting. In this setting, parties to an agreed-upon
workflow do not trust each other to follow that workflow,
or suspect the other party might misrepresent proceedings
at a later time. We demonstrate how distributed declara-
tive workflow execution can be implemented as smart con-
tracts, guaranteeing (I) enforcement of workflow semantics,
and (II) an incontrovertible record of workflow execution his-
tory. Crucially, we achieve both properties without relying
on a trusted third party.

The implementation is based on the Ethereum blockchain,
inheriting the security properties (I) and (II) from the guar-
antees given by that chain. A recurring challenge for both
the implementation and the analysis is the cost of opera-
tions on Ethereum: This cost must be minimised for honest
parties, and an adversary must be prevented from inflicting
extra cost on others.