#416 Improvement: propertybased tests for circuits fuzz witnesses aga…

[Kappa16]

CLOSE #416

What does this PR do?

Adds property-based fuzzing tests for all Noir circuits to catch constraint gaps that fixed example tests miss. Introduces a Node.js fuzzing harness (circuits/scripts/fuzz_circuits.js) that generates randomized valid/invalid witnesses — computes real Poseidon2 commitments, signs with secp256k1 ECDSA, and runs nargo execute to assert pass/fail outcomes across 8 credential circuits. Adds 86 new Noir-native boundary tests (126 total, up from 40) covering threshold off-by-one, date boundaries, tampered commitments, salt mismatches, and commitment collision resistance. Integrates both nargo test and the fuzz harness into CI as a new circuit-fuzz job.

Closes #

Type of change

• Bug fix
• New feature / credential type
• Refactor / cleanup
• Docs
• CI / tooling

Merge requirements

• CI is green — cargo test (contracts), pnpm tsc --noEmit (frontend), pnpm build (frontend), circuit tests — all green
• Greptile confidence ≥ 4/5 — all review comments addressed, no unresolved threads
• Circuit changes: fixtures/<type>/ artifacts updated
• No NEXT_PUBLIC_ prefix on server-only env vars
• No identity fields stored or logged after KYC provider call
• prehash:false preserved on any issuer signing path touched
• Issuer private key never referenced from client-bundled code

✅ Merge requirements

• All CI checks pass (contracts / frontend / circuits as applicable)
• Every Greptile review comment is addressed — no unresolved review threads
• Greptile confidence score is 4/5 or higher

Notes for reviewers

Two-layer testing strategy:

1. Noir-native tests (in each circuit's main.nr) — 86 new #[test] / #[test(should_fail)] cases that exercise constraint logic directly without the expensive ECDSA overhead. These run fast via nargo test and cover boundary conditions systematically: threshold exact, ±1, zero, max, date epoch, large values, commitment collision resistance, salt binding.

2. External fuzzing harness (circuits/scripts/fuzz_circuits.js, 1,267 lines) — generates random witnesses using crypto.randomBytes, computes Poseidon2 commitments locally (same round constants as merkle_tree.js), signs with the demo issuer key via sign.js, writes Prover.toml, and runs nargo execute end-to-end. Covers pass space (valid commitment + signature + satisfied constraint) and fail space (tampered commitment, wrong salt, threshold violations). Backs up and restores Prover.toml files so the fuzz run never corrupts committed fixtures.

Key design decisions:

• The fuzz harness computes commitments in JS rather than invoking nargo execute on the commit circuit — this reuses the Poseidon2 implementation already in merkle_tree.js and avoids a round-trip to the Noir compiler per test case.
• --iterations N flag controls random test count (default 20); CI runs with --iterations 10 to stay within time budget.
• --circuit NAME flag targets a single circuit for local debugging.
• The set_membership circuit has a pre-existing test failure (test_merkle_path_produces_correct_root) on main — confirmed via git stash + nargo test on clean branch. Not introduced by this PR.

Circuits covered by fuzz harness: age_proof, income_proof, funds_proof, accreditation_proof, kyc_proof, jurisdiction_proof, range_proof, employment_proof.

Circuits covered by Noir boundary tests: all of the above + commit, commit3, aggregate_proof (11 total, 126 tests passing).

[drips-wave]

@Kappa16 Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits