daml-verify

Warning

This software is experimental and not intended for production use. Use at your own risk.

Lightweight formal verification for DAML contracts using the Z3 SMT solver. Proves that critical invariants hold for all possible inputs — not just the ones you thought to test.

Properties

daml-verify ships with 9 properties across three categories:

ID	Property	What it proves
C1	Conservation total	Total output equals total input across transfer paths
C2	Receiver amount	Receiver gets exactly the requested amount
C3	Sender change	Sender change equals totalInput - requested
D1	scaleFees safety	No division-by-zero when amuletPrice > 0
D2	Issuance safety	No division-by-zero when capPerCoupon > 0 and totalCoupons > 0
D3	Ensure sufficient	ensure amount > 0.0 guards all division sites
T1	Transfer temporal	requestedAt < executeBefore follows from preconditions
T2	Allocation temporal	Full temporal chain is consistent
T3	Lock expiry	Lock is always active at creation time

Quick start

# Clone and set up
git clone https://github.com/OpenZeppelin/daml-verify.git && cd daml-verify
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

# Run all 9 proofs
python main.py

Expected output:

daml-verify: 9 properties, 9 proved, 0 disproved

  [PROVED] C1: conservation total
  [PROVED] C2: receiver amount
  [PROVED] C3: sender change
  [PROVED] D1: scaleFees safety
  [PROVED] D2: issuance safety
  [PROVED] D3: ensure sufficient
  [PROVED] T1: transfer temporal
  [PROVED] T2: allocation temporal
  [PROVED] T3: lock expiry

Usage

# Run all properties
python main.py

# Run a single property class
python main.py --class conservation   # C1, C2, C3
python main.py --class division       # D1, D2, D3
python main.py --class temporal       # T1, T2, T3

# Run a single property
python main.py C1

# Run tests
pytest

The exit code is 0 if all proofs pass, 1 if any fail.

How it works

Each property is expressed as a precondition/goal pair over symbolic variables (Z3 reals or integers). The prover checks whether preconditions AND NOT goal is satisfiable:

• UNSAT — the property holds for all inputs (proved).
• SAT — a counterexample exists (disproved), and the model is printed.
• UNKNOWN — the solver could not decide within its limits.

The symbolic models in daml_verify/model/ mirror the logic of the corresponding DAML contract functions. Properties in daml_verify/props/ state what should be true about those models.

Project structure

daml_verify/
  prover.py          # Z3 solver integration
  reporter.py        # Output formatting
  model/
    transfer.py      # CNTS transfer logic
    allocation.py    # CNTS allocation logic
    fees.py          # Splice fee computation
  props/
    conservation.py  # Conservation invariants (C1-C3)
    division.py      # Division safety (D1-D3)
    temporal.py      # Temporal ordering (T1-T3)
tests/               # pytest suite (one test per property)
main.py              # CLI entry point

Adding your own properties

1. Model your DAML contract logic in daml_verify/model/ using Z3 symbolic variables.
2. Define a property in daml_verify/props/ as a function returning (preconditions, goal).
3. Register it in ALL_PROPERTIES in daml_verify/prover.py.
4. Add a test in tests/ that asserts ProofStatus.PROVED.

Example property:

from z3 import Reals, And

def prop_my_invariant():
    x, y = Reals("x y")
    preconditions = And(x > 0, y > 0)
    goal = x + y > 0
    return preconditions, goal

Requirements

• Python 3.10+
• z3-solver >= 4.12