diff --git a/.github/workflows/fuzz-afl.yml b/.github/workflows/fuzz-afl.yml
index ce019cc3d..3a2b619b7 100644
--- a/.github/workflows/fuzz-afl.yml
+++ b/.github/workflows/fuzz-afl.yml
@@ -21,36 +21,13 @@ jobs:
     outputs:
       targets: ${{ steps.list.outputs.targets }}
     steps:
-      - name: Build target list
+      - name: Checkout repository
+        uses: actions/checkout@v4
+      # Derive the matrix straight from fuzz/Cargo.toml (the single source of truth),
+      # so new [[bin]] targets are picked up with no manual list to maintain here.
+      - name: Resolve fuzz target matrix
         id: list
-        run: |
-          # Canonical, human-readable list (one target per line) → compact JSON array.
-          targets=$(jq -R -s -c 'split("\n") | map(select(length > 0))' <<'EOF'
-          fuzz_apply_state_diff_split_path
-          fuzz_block_verification
-          fuzz_encoding_roundtrip
-          fuzz_multi_block_state_sequence
-          fuzz_program_deployment_lifecycle
-          fuzz_replay_prevention
-          fuzz_sequencer_vs_replayer
-          fuzz_signature_verification
-          fuzz_state_diff_computation
-          fuzz_state_serialization
-          fuzz_state_transition
-          fuzz_stateless_verification
-          fuzz_transaction_decoding
-          fuzz_validate_execute_consistency
-          fuzz_witness_set_verification
-          fuzz_merkle_tree
-          fuzz_transaction_properties
-          fuzz_privacy_preserving_witness
-          fuzz_encoding_privacy_preserving
-          fuzz_nullifier_set_roundtrip
-          fuzz_privacy_preserving_state_transition
-          EOF
-          )
-          echo "targets=$targets" >> "$GITHUB_OUTPUT"
-          echo "Resolved ${targets}"
+        uses: ./.github/actions/resolve-targets
 
   # ────────────────────────────────────────────────────────────────────────────
   # afl-smoke — 60-second per targets
diff --git a/.github/workflows/fuzz.yml b/.github/workflows/fuzz.yml
index 94e8cf11e..26e25fa31 100644
--- a/.github/workflows/fuzz.yml
+++ b/.github/workflows/fuzz.yml
@@ -20,35 +20,26 @@ jobs:
       - uses: actions/checkout@v4
       - run: python3 scripts/check_target_inventory.py
 
+  # ── Resolve the fuzz target matrix from fuzz/Cargo.toml (single source of truth) ─
+  resolve:
+    name: Resolve target matrix
+    runs-on: ubuntu-latest
+    outputs:
+      targets: ${{ steps.list.outputs.targets }}
+    steps:
+      - uses: actions/checkout@v4
+      - id: list
+        uses: ./.github/actions/resolve-targets
+
   # ── Smoke fuzz: 60 s per target ─────────────────────────────────────────────
   smoke-fuzz:
     name: Smoke fuzz (${{ matrix.target }})
     runs-on: ubuntu-latest
+    needs: resolve
     strategy:
       fail-fast: false
       matrix:
-        target:
-          - fuzz_transaction_decoding
-          - fuzz_stateless_verification
-          - fuzz_state_transition
-          - fuzz_block_verification
-          - fuzz_encoding_roundtrip
-          - fuzz_signature_verification
-          - fuzz_replay_prevention
-          - fuzz_state_diff_computation
-          - fuzz_validate_execute_consistency
-          - fuzz_state_serialization
-          - fuzz_witness_set_verification
-          - fuzz_program_deployment_lifecycle
-          - fuzz_apply_state_diff_split_path
-          - fuzz_multi_block_state_sequence
-          - fuzz_sequencer_vs_replayer
-          - fuzz_merkle_tree
-          - fuzz_transaction_properties
-          - fuzz_privacy_preserving_witness
-          - fuzz_encoding_privacy_preserving
-          - fuzz_nullifier_set_roundtrip
-          - fuzz_privacy_preserving_state_transition
+        target: ${{ fromJSON(needs.resolve.outputs.targets) }}
     steps:
       - uses: actions/checkout@v4
 
@@ -202,31 +193,11 @@ jobs:
   regression:
     name: Corpus regression (${{ matrix.target }})
     runs-on: ubuntu-latest
+    needs: resolve
     strategy:
       fail-fast: false
       matrix:
-        target:
-          - fuzz_transaction_decoding
-          - fuzz_stateless_verification
-          - fuzz_state_transition
-          - fuzz_block_verification
-          - fuzz_encoding_roundtrip
-          - fuzz_signature_verification
-          - fuzz_replay_prevention
-          - fuzz_state_diff_computation
-          - fuzz_validate_execute_consistency
-          - fuzz_state_serialization
-          - fuzz_witness_set_verification
-          - fuzz_program_deployment_lifecycle
-          - fuzz_apply_state_diff_split_path
-          - fuzz_multi_block_state_sequence
-          - fuzz_sequencer_vs_replayer
-          - fuzz_merkle_tree
-          - fuzz_transaction_properties
-          - fuzz_privacy_preserving_witness
-          - fuzz_encoding_privacy_preserving
-          - fuzz_nullifier_set_roundtrip
-          - fuzz_privacy_preserving_state_transition
+        target: ${{ fromJSON(needs.resolve.outputs.targets) }}
     steps:
       - uses: actions/checkout@v4
       - name: Checkout logos-execution-zone
@@ -255,6 +226,7 @@ jobs:
   perf-baseline:
     name: Performance baseline
     runs-on: ubuntu-latest
+    needs: resolve
     if: github.event_name == 'schedule'
     steps:
       - uses: actions/checkout@v4
@@ -263,28 +235,7 @@ jobs:
       - uses: ./.github/actions/setup-libfuzzer
       - name: Measure throughput (30 s per target)
         run: |
-          for target in \
-            fuzz_transaction_decoding \
-            fuzz_stateless_verification \
-            fuzz_state_transition \
-            fuzz_block_verification \
-            fuzz_encoding_roundtrip \
-            fuzz_signature_verification \
-            fuzz_replay_prevention \
-            fuzz_state_diff_computation \
-            fuzz_validate_execute_consistency \
-            fuzz_state_serialization \
-            fuzz_witness_set_verification \
-            fuzz_program_deployment_lifecycle \
-            fuzz_apply_state_diff_split_path \
-            fuzz_multi_block_state_sequence \
-            fuzz_sequencer_vs_replayer \
-            fuzz_merkle_tree \
-            fuzz_transaction_properties \
-            fuzz_privacy_preserving_witness \
-            fuzz_encoding_privacy_preserving \
-            fuzz_nullifier_set_roundtrip \
-            fuzz_privacy_preserving_state_transition; do
+          for target in $(jq -r '.[]' <<< '${{ needs.resolve.outputs.targets }}'); do
             echo "=== $target ===" | tee -a perf_baseline.txt
             cargo fuzz run "$target" -- -max_total_time=30 2>&1 \
               | grep -E "exec/s|execs_per_sec" | tail -1 | tee -a perf_baseline.txt
diff --git a/.github/workflows/mutants.yml b/.github/workflows/mutants.yml
index 829503a34..825fff561 100644
--- a/.github/workflows/mutants.yml
+++ b/.github/workflows/mutants.yml
@@ -139,23 +139,18 @@ jobs:
       - name: Make corpus-regression wrapper executable
         run: chmod +x scripts/mutants-corpus-test.sh
 
-      # Build all 20 fuzz targets once before the mutation loop so that each
-      # mutant only needs to rebuild the mutated crate, not the fuzz harness.
-      # Keep this list in sync with scripts/mutants-corpus-test.sh.
+      # Resolve the target list from fuzz/Cargo.toml (the single source of truth)
+      # so this workflow needs no hand-maintained list; scripts/mutants-corpus-test.sh
+      # parses the same file, so the two stay in sync automatically.
+      - name: Resolve fuzz target matrix
+        id: list
+        uses: ./.github/actions/resolve-targets
+
+      # Build every fuzz target once before the mutation loop so that each mutant
+      # only needs to rebuild the mutated crate, not the fuzz harness.
       - name: Pre-build fuzz targets
         run: |
-          for target in \
-            fuzz_transaction_decoding fuzz_stateless_verification \
-            fuzz_state_transition fuzz_block_verification \
-            fuzz_encoding_roundtrip fuzz_signature_verification \
-            fuzz_replay_prevention fuzz_state_diff_computation \
-            fuzz_validate_execute_consistency fuzz_state_serialization \
-            fuzz_witness_set_verification fuzz_program_deployment_lifecycle \
-            fuzz_apply_state_diff_split_path fuzz_multi_block_state_sequence \
-            fuzz_sequencer_vs_replayer fuzz_merkle_tree \
-            fuzz_transaction_properties fuzz_privacy_preserving_witness \
-            fuzz_encoding_privacy_preserving fuzz_nullifier_set_roundtrip \
-            fuzz_privacy_preserving_state_transition; do
+          for target in $(jq -r '.[]' <<< '${{ steps.list.outputs.targets }}'); do
             cargo fuzz build "${target}"
           done
 
diff --git a/README.md b/README.md
index d2544640e..935b715fd 100644
--- a/README.md
+++ b/README.md
@@ -6,7 +6,7 @@
 [Logos Execution Zone (LEZ)](https://github.com/logos-blockchain/logos-execution-zone) protocol.**
 
 [![Rust](https://img.shields.io/badge/rust-nightly-orange?logo=rust)](rust-toolchain.toml)
-[![Fuzzing](https://img.shields.io/badge/libFuzzer%20%C2%B7%20AFL%2B%2B-21%20targets-blue)](#-fuzz-targets)
+[![Fuzzing](https://img.shields.io/badge/libFuzzer%20%C2%B7%20AFL%2B%2B-22%20targets-blue)](#-fuzz-targets)
 [![Mutation testing](https://img.shields.io/badge/cargo--mutants-enabled-green)](.github/workflows/mutants.yml)
 [![License](https://img.shields.io/badge/license-MIT-lightgrey)](LICENSE-MIT)
 
@@ -31,7 +31,7 @@ lez-fuzzing/
 │       └── generators.rs       # Arbitrary / proptest strategies
 ├── fuzz/                   # cargo-fuzz crate (own [workspace] sentinel)
 │   ├── Cargo.toml
-│   ├── fuzz_targets/       # 21 targets total — see table below
+│   ├── fuzz_targets/       # 22 targets total — see table below
 │   │   ├── _template.rs    # Template for `just new-target`
 │   │   └── fuzz_*.rs
 │   └── corpus/             # Curated seed inputs (one dir per target)
@@ -131,6 +131,7 @@ just fuzz-props
 | 19 | `fuzz_encoding_privacy_preserving` | Privacy-preserving encoding: MessageEncodingRoundtrip + TxEncodingDeterministic/NonEmpty |
 | 20 | `fuzz_nullifier_set_roundtrip` | `NullifierSet` Borsh serialisation: NullifierSetRoundtrip (decode→encode identity for the hand-written impl) |
 | 21 | `fuzz_privacy_preserving_state_transition` | Path B — `NSSATransaction::PrivacyPreserving` through `execute_check_on_state` with a dev-mode passing proof: reaches commitment/nullifier checks 5–6 + `apply_state_diff`. Asserts no-panic, StateIsolationOnFailure, PrivateStateIsolationOnFailure, CommitmentInsertion, NonceIncrementCorrectness, PostStateApplied, ReplayRejection (balance conservation intentionally not asserted — the fake proof bypasses the circuit guarantee) |
+| 22 | `fuzz_transaction_ordering_independence` | Transaction ordering-independence on the shielded path: builds a *nullifier-conflicting* pair (two distinct privacy-preserving txs declaring the same nullifier) and applies it in both orders on independent clones of a seeded state, at an identical `(block_id, timestamp)`. Asserts **NoDoubleSpend** (neither ordering accepts both — the shared nullifier is spendable at most once) and **OrderIndependentAcceptance** (the count of accepted txs is the same in both orderings). The nullifier check is enforced by the state machine, not the circuit, so the dev-mode fake proof does not mask it. Requires `RISC0_DEV_MODE=1` |
 
 Each target lives at `fuzz/fuzz_targets/<name>.rs`.
 
@@ -167,13 +168,19 @@ cp fuzz/artifacts/fuzz_state_transition/crash-abc123-minimised \
 ## ➕ Adding a New Target
 
 ```bash
-# Scaffold everything automatically (corpus dir, .rs file, Cargo.toml entry, CI matrix entry)
+# Scaffold everything automatically (corpus dir, .rs file, Cargo.toml entry)
 just new-target my_feature   # creates fuzz_my_feature
 ```
 
 `just new-target` calls [`scripts/add_fuzz_target.py`](scripts/add_fuzz_target.py), which
-appends the `[[bin]]` entry to [`fuzz/Cargo.toml`](fuzz/Cargo.toml) and inserts the target
-into every strategy matrix in [`.github/workflows/fuzz.yml`](.github/workflows/fuzz.yml).
+appends the `[[bin]]` entry to [`fuzz/Cargo.toml`](fuzz/Cargo.toml) — the **single source of
+truth**. Every workflow and script derives its target list from that file at runtime (the CI
+matrices and build loops via the [`resolve-targets`](.github/actions/resolve-targets)
+composite action, and [`scripts/mutants-corpus-test.sh`](scripts/mutants-corpus-test.sh) via
+an inline parse), so **no CI edits are needed**. The only manual step is a prose row in the
+target tables of `README.md` and [`docs/fuzzing.md`](docs/fuzzing.md);
+[`scripts/check_target_inventory.py`](scripts/check_target_inventory.py) (run in CI) fails the
+build if either table drifts from `fuzz/Cargo.toml`.
 
 ---
 
diff --git a/corpus/libfuzz/fuzz_state_transition/seed_empty_tx b/corpus/libfuzz/fuzz_state_transition/seed_empty_tx
index b00ba8713..61f4c0a63 100644
Binary files a/corpus/libfuzz/fuzz_state_transition/seed_empty_tx and b/corpus/libfuzz/fuzz_state_transition/seed_empty_tx differ
diff --git a/corpus/libfuzz/fuzz_stateless_verification/seed_empty_tx b/corpus/libfuzz/fuzz_stateless_verification/seed_empty_tx
index b00ba8713..61f4c0a63 100644
Binary files a/corpus/libfuzz/fuzz_stateless_verification/seed_empty_tx and b/corpus/libfuzz/fuzz_stateless_verification/seed_empty_tx differ
diff --git a/corpus/libfuzz/fuzz_transaction_decoding/seed_empty_tx b/corpus/libfuzz/fuzz_transaction_decoding/seed_empty_tx
index b00ba8713..61f4c0a63 100644
Binary files a/corpus/libfuzz/fuzz_transaction_decoding/seed_empty_tx and b/corpus/libfuzz/fuzz_transaction_decoding/seed_empty_tx differ
diff --git a/current_vs_alternative_approach.md b/current_vs_alternative_approach.md
index f4923662f..9d43c34e6 100644
--- a/current_vs_alternative_approach.md
+++ b/current_vs_alternative_approach.md
@@ -11,7 +11,7 @@ The `lez-fuzzing` repository is a **coverage-guided, structured mutation fuzzing
 | Rich generators | [`fuzz_props::generators`](fuzz_props/src/generators.rs) adds `proptest` strategies for pathological sequences, phantom-account attacks, overflow amounts, replay sequences |
 | Protocol invariants | [`fuzz_props::invariants`](fuzz_props/src/invariants.rs) expresses zero-mutation-on-rejection and replay-rejection as reusable `ProtocolInvariant` objects |
 | ZK-awareness | `RISC0_DEV_MODE=1` stubs out `risc0-zkvm` proofs, enabling ~5 000–200 000 exec/sec depending on target |
-| 20 dedicated targets | Covers encoding, signature verification, stateless checks, state transitions, state diffs, replay prevention, validate/execute consistency, block verification, state serialization, witness-set verification, program deployment lifecycle, split-path equivalence, multi-block sequences, sequencer-vs-replayer differential, Merkle-tree invariants, transaction properties, privacy-preserving witness/encoding, and nullifier-set round-trips. Input-independent invariant checks (genesis contents, getters, system-account guard) are kept as **LEZ unit tests**, not targets — see [`docs/mutants-not-fuzzable.md`](docs/mutants-not-fuzzable.md) |
+| 22 dedicated targets | Covers encoding, signature verification, stateless checks, state transitions, state diffs, replay prevention, validate/execute consistency, block verification, state serialization, witness-set verification, program deployment lifecycle, split-path equivalence, multi-block sequences, sequencer-vs-replayer differential, Merkle-tree invariants, transaction properties, privacy-preserving witness/encoding, nullifier-set round-trips, the privacy-preserving state-transition executor, and transaction ordering-independence (shielded-path nullifier double-spend across orderings). Input-independent invariant checks (genesis contents, getters, system-account guard) are kept as **LEZ unit tests**, not targets — see [`docs/mutants-not-fuzzable.md`](docs/mutants-not-fuzzable.md) |
 | CI integration | GitHub Actions libFuzzer (`fuzz.yml`), AFL++ (`fuzz-afl.yml`), and mutation-testing (`mutants.yml`) workflows run on every PR / nightly |
 | Pre-seeded corpus | Hundreds of minimised seed files in [`fuzz/corpus/`](fuzz/corpus/) ensure regressions are caught instantly |
 
diff --git a/docs/fuzzing.md b/docs/fuzzing.md
index b1f95e7da..de9e74141 100644
--- a/docs/fuzzing.md
+++ b/docs/fuzzing.md
@@ -130,6 +130,7 @@ just fuzz-regression
 | `fuzz_encoding_privacy_preserving` | Privacy-preserving encoding: **MessageEncodingRoundtrip**, **TxEncodingDeterministic** / **NonEmpty** | `fuzz/fuzz_targets/fuzz_encoding_privacy_preserving.rs` |
 | `fuzz_nullifier_set_roundtrip` | `NullifierSet` Borsh serialisation: **NullifierSetRoundtrip** (decode→encode identity for the hand-written impl) | `fuzz/fuzz_targets/fuzz_nullifier_set_roundtrip.rs` |
 | `fuzz_privacy_preserving_state_transition` | Path B — `NSSATransaction::PrivacyPreserving` through `execute_check_on_state` with a dev-mode passing proof, reaching checks 5–6 (`check_commitments_are_new` / `check_nullifiers_are_valid`) and `apply_state_diff`: **No panic**, **StateIsolationOnFailure**, **PrivateStateIsolationOnFailure**, **CommitmentInsertion**, **NonceIncrementCorrectness**, **PostStateApplied**, **ReplayRejection**. Balance conservation is intentionally *not* asserted — the synthesised fake receipt bypasses the circuit guarantee. Requires `RISC0_DEV_MODE=1` | `fuzz/fuzz_targets/fuzz_privacy_preserving_state_transition.rs` |
+| `fuzz_transaction_ordering_independence` | Ordering-independence for the shielded path — the only target that compares two *orderings* of the same transactions rather than one fixed order. `arb_conflicting_nullifier_pair` builds two distinct privacy-preserving txs declaring the **same** nullifier; they are applied in both orders (`B→C` and `C→B`) on independent clones of a commitment-seeded state, at an identical `(block_id, timestamp)` so order is the only variable. Asserts **NoDoubleSpend** (a shared nullifier is spendable at most once per ordering) and **OrderIndependentAcceptance** (the number of accepted txs is order-invariant). The nullifier guard is a state-machine check, not a circuit check, so the dev-mode fake receipt does not mask a violation. Requires `RISC0_DEV_MODE=1` | `fuzz/fuzz_targets/fuzz_transaction_ordering_independence.rs` |
 
 ---
 
@@ -147,8 +148,7 @@ This single command does four things automatically:
 |---|---|
 | Creates the corpus directory | `fuzz/corpus/fuzz_my_feature/` |
 | Writes a typed fuzz target template | `fuzz/fuzz_targets/fuzz_my_feature.rs` |
-| Appends `[[bin]]` entry to `fuzz/Cargo.toml` | Covers **both** the libFuzzer and AFL++ lanes |
-| Inserts target into every CI matrix + perf loop | `.github/workflows/fuzz.yml` |
+| Appends `[[bin]]` entry to `fuzz/Cargo.toml` | Covers **both** the libFuzzer and AFL++ lanes — and every workflow/script, which derive their target lists from this file |
 
 The generated template uses `fuzz_props::fuzz_entry!` and works with both engines
 without modification.
@@ -164,16 +164,26 @@ structured input, or the proptest generators from
 ### Step 3 — Automated registration (cargo-fuzz + CI)
 
 `just new-target` calls [`scripts/add_fuzz_target.py`](../scripts/add_fuzz_target.py)
-which:
-- Appends the `[[bin]]` entry to [`fuzz/Cargo.toml`](../fuzz/Cargo.toml).
-  This **single entry** covers both the libFuzzer lane (`cargo fuzz build`) and
-  the AFL++ lane (`cargo afl build --no-default-features --features fuzzer-afl`).
-- Inserts the target name into every strategy matrix and the perf-baseline shell
-  loop in [`.github/workflows/fuzz.yml`](../.github/workflows/fuzz.yml).
+which appends the `[[bin]]` entry to [`fuzz/Cargo.toml`](../fuzz/Cargo.toml). This
+**single entry** covers both the libFuzzer lane (`cargo fuzz build`) and the AFL++
+lane (`cargo afl build --no-default-features --features fuzzer-afl`) — and it is the
+**single source of truth** every workflow and script reads at runtime:
+
+- The CI matrices and build loops (`fuzz.yml`, `fuzz-afl.yml`, `corpus-update.yml`,
+  `mutants.yml`) resolve their target list through the
+  [`resolve-targets`](../.github/actions/resolve-targets) composite action, which
+  parses `fuzz/Cargo.toml`.
+- [`scripts/mutants-corpus-test.sh`](../scripts/mutants-corpus-test.sh) parses the
+  same file inline.
+
+So a new `[[bin]]` needs **no workflow edits** — the only hand-authored places are the
+prose target tables in this file and `README.md`, which
+[`scripts/check_target_inventory.py`](../scripts/check_target_inventory.py) (run in CI)
+checks against `fuzz/Cargo.toml`.
 
 > [!TIP]
 > **Manual fallback:** if you create a target without `just new-target`, add the
-> entry yourself:
+> entry yourself — that alone wires it into every lane:
 >
 > ```toml
 > [[bin]]
@@ -206,7 +216,8 @@ cd fuzz && cargo afl build \
 | `fuzz/corpus/fuzz_<name>/` | Create | ✅ `just new-target` |
 | `fuzz/Cargo.toml` | Add `[[bin]]` (covers both lanes) | ✅ `just new-target` |
 | `Justfile` | Nothing — auto-discovers | ✅ automatic |
-| `.github/workflows/fuzz.yml` | Add to 3 matrix lists | ✅ `just new-target` |
+| `.github/workflows/*.yml`, `scripts/mutants-corpus-test.sh` | Nothing — target lists derive from `fuzz/Cargo.toml` | ✅ automatic |
+| `README.md`, `docs/fuzzing.md` | Add a prose row to the target table | ⚠️ manual (CI-gated by `check_target_inventory.py`) |
 
 ---
 
@@ -377,8 +388,8 @@ The nightly AFL++ CI workflow has two jobs:
 
 | Job | Triggers | Matrix |
 |-----|----------|--------|
-| `afl-smoke` | nightly + `workflow_dispatch` | all 21 targets, 60 s each |
-| `afl-coverage-aggregate` | nightly, `needs: afl-smoke` | all 21 targets merged into one LLVM HTML report |
+| `afl-smoke` | nightly + `workflow_dispatch` | all 22 targets, 60 s each |
+| `afl-coverage-aggregate` | nightly, `needs: afl-smoke` | all 22 targets merged into one LLVM HTML report |
 
 The smoke job (one matrix leg per target, on `ubuntu-latest`):
 1. Builds AFL++ from source, then builds the target with `cargo afl build --no-default-features --features fuzzer-afl`
@@ -388,7 +399,7 @@ The smoke job (one matrix leg per target, on `ubuntu-latest`):
 
 The coverage-aggregate job:
 1. Downloads every smoke leg's findings
-2. Rebuilds all 21 targets with `RUSTFLAGS="-C instrument-coverage"`
+2. Rebuilds all 22 targets with `RUSTFLAGS="-C instrument-coverage"`
 3. Runs all checked-in corpus + AFL queue inputs through each binary
 4. Merges every `.profraw` → one `.profdata` → a single combined HTML report via `llvm-cov show`
 
@@ -622,6 +633,7 @@ Measured on a 4-core x86_64 Linux runner with `RISC0_DEV_MODE=1`:
 | `fuzz_encoding_privacy_preserving` | ~50 000 exec/sec *(estimate)* |
 | `fuzz_nullifier_set_roundtrip` | ~100 000 exec/sec *(estimate)* |
 | `fuzz_privacy_preserving_state_transition` | slow — dev-mode proof synthesis + verification per exec dominates runtime *(estimate)* |
+| `fuzz_transaction_ordering_independence` | slow — seeds the commitment set, then synthesises proofs and executes the conflicting pair in both orderings per exec *(estimate)* |
 
 > [!NOTE]
 > Throughput figures for the five new targets are rough estimates; run `just perf-baseline`
diff --git a/fuzz/Cargo.toml b/fuzz/Cargo.toml
index ef906cdea..a138fde00 100644
--- a/fuzz/Cargo.toml
+++ b/fuzz/Cargo.toml
@@ -157,3 +157,9 @@ name = "fuzz_privacy_preserving_state_transition"
 path = "fuzz_targets/fuzz_privacy_preserving_state_transition.rs"
 test = false
 bench = false
+
+[[bin]]
+name = "fuzz_transaction_ordering_independence"
+path = "fuzz_targets/fuzz_transaction_ordering_independence.rs"
+test = false
+bench = false
diff --git a/fuzz/fuzz_targets/fuzz_transaction_ordering_independence.rs b/fuzz/fuzz_targets/fuzz_transaction_ordering_independence.rs
new file mode 100644
index 000000000..9504b8059
--- /dev/null
+++ b/fuzz/fuzz_targets/fuzz_transaction_ordering_independence.rs
@@ -0,0 +1,133 @@
+#![cfg_attr(feature = "fuzzer-libfuzzer", no_main)]
+//! Fuzz target: transaction **ordering-independence** for the shielded (privacy-preserving)
+//! path — the property no other target asserts.
+//!
+//! Every other target applies transactions in a single fixed order and checks the *result*.
+//! None asks whether the *order* of two transactions can change which ones are accepted. This
+//! target does: it applies a nullifier-conflicting pair in both orders on independent copies
+//! of the same base state and asserts the outcome is order-independent.
+//!
+//! # Why the shielded path (and not native transfers)
+//!
+//! The original `fuzz_transaction_non_interference` proposal aimed this idea at native
+//! transfers, but `V03State`'s only cross-transaction state is the append-only
+//! `(CommitmentSet, NullifierSet)`; two disjoint public transfers touch only per-account maps
+//! and commute trivially, so that target would be permanently green. The nullifier set is the
+//! real shared state — it is what prevents double-spends — so that is where ordering can
+//! actually matter.
+//!
+//! # The oracle
+//!
+//! `arb_conflicting_nullifier_pair` builds two *distinct* transactions `B` and `C` that
+//! declare the **same** nullifier. A correct state machine enforces first-come-first-served:
+//! whichever is applied first spends the nullifier, and the second is rejected at
+//! `check_nullifiers_are_valid`. Both orderings are applied at an **identical** `(block_id,
+//! timestamp)` so the *only* difference between them is transaction order — fixing a flaw in
+//! the original sketch, which varied the clock per position and so could not distinguish
+//! ordering effects from validity-window effects.
+//!
+//! Note on scope: this nullifier check is enforced by the *state machine*, not by the ZK
+//! circuit, so the dev-mode synthesised proof (which bypasses the circuit) does **not** mask
+//! it — unlike balance conservation, which this path deliberately never asserts.
+//!
+//! Requires `RISC0_DEV_MODE=1` (set by every `just fuzz` recipe) for the synthesised proofs.
+//!
+//! # Invariants asserted
+//!
+//! * **NoDoubleSpend** — in neither ordering are both conflicting transactions accepted; a
+//!   shared nullifier is spendable at most once. A violation is a literal double-spend.
+//! * **OrderIndependentAcceptance** — the number of transactions accepted from the pair is the
+//!   same in both orderings. A violation means acceptance leaks across transactions depending
+//!   on order (order-dependent interference through global state).
+
+use arbitrary::{Arbitrary, Unstructured};
+use common::transaction::LeeTransaction;
+use fuzz_props::generators::arbitrary_fuzz_state;
+use fuzz_props::privacy::{arb_conflicting_nullifier_pair, arb_privacy_preserving_tx};
+use nssa::{AccountId, PrivacyPreservingTransaction};
+
+/// Apply the production stateless gate and wrap for execution; `None` drops the input.
+fn gate(tx: PrivacyPreservingTransaction) -> Option<LeeTransaction> {
+    LeeTransaction::PrivacyPreserving(tx)
+        .transaction_stateless_check()
+        .ok()
+}
+
+fuzz_props::fuzz_entry!(|data: &[u8]| {
+    let mut u = Unstructured::new(data);
+
+    // Need at least two keyed accounts so the conflicting pair can use distinct signers.
+    let fuzz_accs = match arbitrary_fuzz_state(&mut u) {
+        Ok(accs) if accs.len() >= 2 => accs,
+        _ => return,
+    };
+    let init_accs: Vec<(AccountId, u128)> = fuzz_accs
+        .iter()
+        .map(|a| (a.account_id, a.balance))
+        .collect();
+    let mut base = fuzz_props::genesis::genesis_state(&init_accs, vec![]);
+
+    // ── Seed the commitment set ──────────────────────────────────────────────────────────
+    // A nullifier only passes check 6 when its digest is in `root_history`, which starts empty
+    // and is seeded only once a commitment-bearing transaction applies. Reuse the
+    // proven-reachable generator to grow it; individual outcomes don't matter here.
+    let n_seed: u8 = u8::arbitrary(&mut u).unwrap_or(0) % 4;
+    for i in 0..n_seed {
+        let Ok(tx) = arb_privacy_preserving_tx(&mut u, &base, &fuzz_accs) else {
+            break;
+        };
+        let Some(lee) = gate(tx) else { continue };
+        let _ = lee.execute_check_on_state(&mut base, 1 + u64::from(i), u64::from(i));
+    }
+
+    // ── Build the nullifier-conflicting pair against the seeded base ─────────────────────
+    let Ok((tx_b, tx_c)) = arb_conflicting_nullifier_pair(&mut u, &base, &fuzz_accs) else {
+        return;
+    };
+
+    // Two independent instances of each so both orderings get a fresh, un-consumed copy.
+    let (Some(b1), Some(c1)) = (gate(tx_b.clone()), gate(tx_c.clone())) else {
+        return;
+    };
+    let (Some(b2), Some(c2)) = (gate(tx_b), gate(tx_c)) else {
+        return;
+    };
+
+    // Identical clock for both orderings: the sole difference is transaction order. The pair's
+    // validity windows are unbounded, so the specific values are immaterial.
+    const BLOCK: u64 = 1;
+    const TS: u64 = 0;
+
+    // ── Order 1: B → C ───────────────────────────────────────────────────────────────────
+    let mut s_bc = base.clone();
+    let rb1 = b1.execute_check_on_state(&mut s_bc, BLOCK, TS).is_ok();
+    let rc1 = c1.execute_check_on_state(&mut s_bc, BLOCK, TS).is_ok();
+
+    // ── Order 2: C → B ───────────────────────────────────────────────────────────────────
+    let mut s_cb = base.clone();
+    let rc2 = c2.execute_check_on_state(&mut s_cb, BLOCK, TS).is_ok();
+    let rb2 = b2.execute_check_on_state(&mut s_cb, BLOCK, TS).is_ok();
+
+    // ── INVARIANT [NoDoubleSpend] ────────────────────────────────────────────────────────
+    // The shared nullifier must be spendable at most once, in either order.
+    assert!(
+        !(rb1 && rc1),
+        "INVARIANT VIOLATION [NoDoubleSpend]: both conflicting transactions accepted in order \
+         B→C — the shared nullifier was double-spent"
+    );
+    assert!(
+        !(rc2 && rb2),
+        "INVARIANT VIOLATION [NoDoubleSpend]: both conflicting transactions accepted in order \
+         C→B — the shared nullifier was double-spent"
+    );
+
+    // ── INVARIANT [OrderIndependentAcceptance] ───────────────────────────────────────────
+    // The count of accepted transactions from the pair must not depend on ordering.
+    let accepted_bc = u8::from(rb1) + u8::from(rc1);
+    let accepted_cb = u8::from(rb2) + u8::from(rc2);
+    assert_eq!(
+        accepted_bc, accepted_cb,
+        "INVARIANT VIOLATION [OrderIndependentAcceptance]: {accepted_bc} of the pair accepted \
+         as B→C but {accepted_cb} as C→B — transaction acceptance is order-dependent",
+    );
+});
diff --git a/fuzz_props/src/privacy.rs b/fuzz_props/src/privacy.rs
index f88923a78..8b1f46594 100644
--- a/fuzz_props/src/privacy.rs
+++ b/fuzz_props/src/privacy.rs
@@ -321,3 +321,89 @@ pub fn arb_privacy_preserving_tx(
     let witness_set = PPWitnessSet::for_message(&message, proof, &keys);
     Ok(PrivacyPreservingTransaction::new(message, witness_set))
 }
+
+/// Build a minimal *pure-private* transaction: one signer, no public accounts, the given
+/// commitments and nullifiers, and unbounded validity windows.
+///
+/// Two properties matter for the ordering-independence oracle in
+/// `fuzz_transaction_ordering_independence`:
+///
+/// * **`public_account_ids` is empty**, so `synthesize_passing_proof` reconstructs an empty
+///   `public_pre_states` and the journal does not depend on live chain state. The proof
+///   therefore stays valid at check 4 even after another transaction has mutated the state —
+///   i.e. it is valid whether this tx is applied *first* or *second*. That is what lets us
+///   apply the same transaction in both orderings and compare outcomes soundly.
+/// * The single signer's nonce is read live from `state`, so check 3c passes by construction
+///   at first application; because the paired transaction uses a *different* signer, this
+///   signer's nonce is untouched when this tx is applied second — so any rejection of the
+///   second application is attributable to the shared nullifier, not to a nonce mismatch.
+fn build_pure_private_tx(
+    state: &V03State,
+    key: &PrivateKey,
+    new_commitments: Vec<Commitment>,
+    new_nullifiers: Vec<(Nullifier, CommitmentSetDigest)>,
+) -> PrivacyPreservingTransaction {
+    let signer_id = account_id_for_key(key);
+    let message = PPMessage {
+        public_account_ids: Vec::new(),
+        nonces: vec![state.get_account_by_id(signer_id).nonce],
+        public_post_states: Vec::new(),
+        encrypted_private_post_states: Vec::new(),
+        new_commitments,
+        new_nullifiers,
+        block_validity_window: ValidityWindow::new_unbounded(),
+        timestamp_validity_window: ValidityWindow::new_unbounded(),
+    };
+    let proof = synthesize_passing_proof(&message, state, &[signer_id]);
+    let witness_set = PPWitnessSet::for_message(&message, proof, &[key]);
+    PrivacyPreservingTransaction::new(message, witness_set)
+}
+
+/// Build a **nullifier-conflicting pair**: two *distinct* privacy-preserving transactions
+/// (different signers, different fresh commitments) that both declare the **same** nullifier.
+///
+/// The shared nullifier's digest is bound to the current commitment-set root
+/// (`state.commitment_set_digest()`); this only satisfies check 6's `root_history` membership
+/// once a commitment-bearing transaction has already grown the set, so callers should seed the
+/// state first (see the target). The two transactions are otherwise independently valid, so a
+/// correct state machine must accept *at most one* of them regardless of application order —
+/// the property the ordering-independence target asserts.
+///
+/// Returns `Err` when there are fewer than two distinct keyed accounts to draw signers from.
+pub fn arb_conflicting_nullifier_pair(
+    u: &mut Unstructured<'_>,
+    state: &V03State,
+    accounts: &[FuzzAccount],
+) -> ArbResult<(PrivacyPreservingTransaction, PrivacyPreservingTransaction)> {
+    if accounts.len() < 2 {
+        return Err(arbitrary::Error::IncorrectFormat);
+    }
+    // Two distinct signer accounts so the pair's nonce checks are independent.
+    let i = (u8::arbitrary(u)? as usize) % accounts.len();
+    let mut j = (u8::arbitrary(u)? as usize) % accounts.len();
+    if j == i {
+        j = (i + 1) % accounts.len();
+    }
+    let key_b = &accounts[i].private_key;
+    let key_c = &accounts[j].private_key;
+    if account_id_for_key(key_b) == account_id_for_key(key_c) {
+        return Err(arbitrary::Error::IncorrectFormat);
+    }
+
+    // One nullifier, shared by both transactions, bound to a historical commitment-set root.
+    let root = state.commitment_set_digest();
+    let null_aid = AccountId::new(<[u8; 32]>::arbitrary(u)?);
+    let shared_nullifiers = vec![(Nullifier::for_account_initialization(&null_aid), root)];
+
+    // Distinct fresh commitments make the two transactions genuinely different (and keep them
+    // from colliding with each other on check 5).
+    let comm_b = Commitment::new(&AccountId::new(<[u8; 32]>::arbitrary(u)?), &arb_account(u)?);
+    let comm_c = Commitment::new(&AccountId::new(<[u8; 32]>::arbitrary(u)?), &arb_account(u)?);
+    if comm_b == comm_c {
+        return Err(arbitrary::Error::IncorrectFormat);
+    }
+
+    let tx_b = build_pure_private_tx(state, key_b, vec![comm_b], shared_nullifiers.clone());
+    let tx_c = build_pure_private_tx(state, key_c, vec![comm_c], shared_nullifiers);
+    Ok((tx_b, tx_c))
+}
diff --git a/fuzz_props/src/tests/privacy.rs b/fuzz_props/src/tests/privacy.rs
index 2bbde6676..d8317595a 100644
--- a/fuzz_props/src/tests/privacy.rs
+++ b/fuzz_props/src/tests/privacy.rs
@@ -1,8 +1,9 @@
 use arbitrary::Unstructured;
 
-use crate::generators::FuzzAccount;
+use crate::generators::{FuzzAccount, account_id_for_key};
 use crate::privacy::{
-    arb_account, arb_privacy_preserving_tx, arb_validity_window, synthesize_passing_proof,
+    arb_account, arb_conflicting_nullifier_pair, arb_privacy_preserving_tx, arb_validity_window,
+    synthesize_passing_proof,
 };
 use nssa::privacy_preserving_transaction::{Message as PPMessage, WitnessSet as PPWitnessSet};
 use nssa::{AccountId, PrivacyPreservingTransaction, PrivateKey};
@@ -478,3 +479,140 @@ fn arb_privacy_preserving_tx_generator_invariants() {
         "garbage-proof rate {garbage}/{oks} is below 1/16 (expected ~1/8)"
     );
 }
+
+// ── arb_conflicting_nullifier_pair ──────────────────────────────────────────────────────
+// The pair builder underpins `fuzz_transaction_ordering_independence`: it must always yield
+// two transactions that use *distinct* signers (so a rejection of the second application is
+// attributable to the shared nullifier, not a nonce clash) and index only within the account
+// set. These tests pin the account-count guard, the collision-repair branch, and the modular
+// index arithmetic.
+
+/// `n` distinct keyed fuzz accounts (`[1; 32]`, `[2; 32]`, …). Distinct nonzero scalars give
+/// distinct keys and therefore distinct key-derived signer ids.
+fn keyed_accounts(n: u8) -> Vec<FuzzAccount> {
+    (1..=n)
+        .map(|i| FuzzAccount {
+            account_id: AccountId::new([i; 32]),
+            balance: 1_000_000,
+            private_key: PrivateKey::try_new([i; 32]).expect("nonzero scalar is a valid key"),
+        })
+        .collect()
+}
+
+/// The key-derived signer ids the validator would recover from a transaction's witness set.
+fn signer_ids(tx: &PrivacyPreservingTransaction) -> Vec<AccountId> {
+    tx.witness_set()
+        .signatures_and_public_keys()
+        .iter()
+        .map(|(_, pk)| AccountId::from(pk))
+        .collect()
+}
+
+/// The builder needs two distinct accounts: fewer than two is always rejected (before any
+/// index arithmetic runs), exactly two always succeeds. This pins the `accounts.len() < 2`
+/// guard against `==` / `>` / `<=` mutations.
+#[test]
+fn arb_conflicting_nullifier_pair_requires_two_accounts() {
+    let accounts = keyed_accounts(2);
+    let genesis: Vec<(AccountId, u128)> =
+        accounts.iter().map(|a| (a.account_id, a.balance)).collect();
+    let state = crate::genesis::genesis_state(&genesis, vec![]);
+
+    let mut rng = Rng::new();
+    let mut buf = vec![0_u8; 512];
+    rng.fill(&mut buf);
+
+    // Zero accounts: rejected by the guard. A guard mutated to `== 2` / `> 2` would fall
+    // through and divide by `accounts.len() == 0`, panicking — also a failure this catches.
+    let mut u0 = Unstructured::new(&buf);
+    assert!(
+        arb_conflicting_nullifier_pair(&mut u0, &state, &[]).is_err(),
+        "an empty account set must be rejected"
+    );
+    // One account: still rejected — there is no room for two distinct signers.
+    let mut u1 = Unstructured::new(&buf);
+    assert!(
+        arb_conflicting_nullifier_pair(&mut u1, &state, &accounts[..1]).is_err(),
+        "a single-account set must be rejected"
+    );
+    // Two accounts: must succeed. A guard mutated to `== 2` / `<= 2` would reject this.
+    let mut u2 = Unstructured::new(&buf);
+    assert!(
+        arb_conflicting_nullifier_pair(&mut u2, &state, &accounts).is_ok(),
+        "two distinct accounts must yield a pair"
+    );
+}
+
+/// When both index bytes select the same account (`i == j`), the repair branch
+/// `j = (i + 1) % len` must pick the *other* account so the pair keeps distinct signers.
+/// Forcing `i == j == 0` over two accounts exercises that branch: the only correct outcome is
+/// signers `{account 0, account 1}`. This pins the `j == i` test, the `(i + 1) % len` repair,
+/// and the two distinctness guards (`== 2` mutations of them all reject this otherwise-valid
+/// input, and the arithmetic mutations either repair to `j == i` again or index out of range).
+#[test]
+fn arb_conflicting_nullifier_pair_repairs_colliding_indices() {
+    let accounts = keyed_accounts(2);
+    let genesis: Vec<(AccountId, u128)> =
+        accounts.iter().map(|a| (a.account_id, a.balance)).collect();
+    let state = crate::genesis::genesis_state(&genesis, vec![]);
+
+    let mut rng = Rng::new();
+    let mut buf = vec![0_u8; 512];
+    rng.fill(&mut buf);
+    // The first two bytes are the `i` and `j` index draws; zero both so `i == j == 0`.
+    buf[0] = 0;
+    buf[1] = 0;
+
+    let mut u = Unstructured::new(&buf);
+    let (tx_b, tx_c) = arb_conflicting_nullifier_pair(&mut u, &state, &accounts)
+        .expect("colliding indices must be repaired into two distinct signers, not rejected");
+
+    let sb = signer_ids(&tx_b);
+    let sc = signer_ids(&tx_c);
+    assert_eq!(sb.len(), 1, "tx_b must carry exactly one signer");
+    assert_eq!(sc.len(), 1, "tx_c must carry exactly one signer");
+    assert_ne!(
+        sb[0], sc[0],
+        "a conflicting pair must use two distinct signers"
+    );
+
+    let known: std::collections::HashSet<AccountId> = accounts
+        .iter()
+        .map(|a| account_id_for_key(&a.private_key))
+        .collect();
+    assert!(
+        known.contains(&sb[0]) && known.contains(&sc[0]),
+        "both signers must be drawn from the account set"
+    );
+}
+
+/// Over many random inputs the index arithmetic must stay within the account slice. A `% len`
+/// mutated to `/ len` or `+ len` computes an out-of-range index and panics on `accounts[i]`;
+/// the modulo keeps every draw in range and the two signers distinct.
+#[test]
+fn arb_conflicting_nullifier_pair_indexes_in_range() {
+    let accounts = keyed_accounts(2);
+    let genesis: Vec<(AccountId, u128)> =
+        accounts.iter().map(|a| (a.account_id, a.balance)).collect();
+    let state = crate::genesis::genesis_state(&genesis, vec![]);
+    let known: std::collections::HashSet<AccountId> = accounts
+        .iter()
+        .map(|a| account_id_for_key(&a.private_key))
+        .collect();
+
+    let mut rng = Rng::new();
+    let mut buf = vec![0_u8; 512];
+    for _ in 0..500_usize {
+        rng.fill(&mut buf);
+        let mut u = Unstructured::new(&buf);
+        let (tx_b, tx_c) = arb_conflicting_nullifier_pair(&mut u, &state, &accounts)
+            .expect("two distinct accounts always yield a pair");
+        let sb = signer_ids(&tx_b);
+        let sc = signer_ids(&tx_c);
+        assert_ne!(sb[0], sc[0], "conflicting-pair signers must be distinct");
+        assert!(
+            known.contains(&sb[0]) && known.contains(&sc[0]),
+            "signers must be drawn from the account set"
+        );
+    }
+}
diff --git a/scripts/add_fuzz_target.py b/scripts/add_fuzz_target.py
index 97a753186..20c85b79d 100644
--- a/scripts/add_fuzz_target.py
+++ b/scripts/add_fuzz_target.py
@@ -1,25 +1,27 @@
 #!/usr/bin/env python3
-"""Fully automates registering a new cargo-fuzz / AFL++ fuzz target.
+"""Register a new cargo-fuzz / AFL++ fuzz target.
 
 Usage:
     python3 scripts/add_fuzz_target.py <TARGET_NAME>
 
 Where TARGET_NAME is the full binary name, e.g. fuzz_my_feature.
 
-Actions performed:
-  1. Appends a [[bin]] entry to fuzz/Cargo.toml (one entry covers BOTH
-     the libFuzzer lane and the AFL++ lane — no separate Cargo.toml needed)
-  2. Inserts TARGET_NAME into every YAML matrix block in
-     .github/workflows/fuzz.yml  (smoke-fuzz, regression)
-  3. Inserts TARGET_NAME into the perf-baseline shell for-loop in
-     .github/workflows/fuzz.yml
+A single `[[bin]]` entry in fuzz/Cargo.toml is the source of truth for BOTH
+engines and for every workflow/script: the CI matrices and build loops derive
+their target lists from fuzz/Cargo.toml at runtime (via the
+`.github/actions/resolve-targets` composite action, or an inline parse in
+`scripts/mutants-corpus-test.sh`). Appending the `[[bin]]` is therefore all this
+script needs to do — no workflow editing.
 
-NOTE: A single fuzz/Cargo.toml is the source of truth for both engines.
   - libFuzzer build:  cargo fuzz build <TARGET>
   - AFL++ build:      cd fuzz && cargo afl build \\
                         --no-default-features --features fuzzer-afl \\
                         --release --bin <TARGET>
 
+The only remaining manual step is the human-authored target tables in README.md
+and docs/fuzzing.md, which carry a prose description per target that cannot be
+auto-generated. `scripts/check_target_inventory.py` (run in CI) guards those.
+
 Run from the repository root.
 """
 
@@ -49,107 +51,6 @@ def append_cargo_bin(target: str, cargo_toml: Path) -> None:
     print(f"  [+] fuzz/Cargo.toml  — added [[bin]] {target!r}")
 
 
-def insert_into_yaml_matrices(target: str, content: str) -> tuple[str, int]:
-    """Insert target into YAML strategy matrix blocks.
-
-    Matches blocks of the form::
-
-        target:
-          - fuzz_a
-          - fuzz_b
-
-    and appends ``          - <target>`` after the last existing entry.
-    """
-    pattern = re.compile(
-        r"(        target:\n(?:          - fuzz_\w+\n)+)",
-        re.MULTILINE,
-    )
-
-    def add_target(m: re.Match) -> str:
-        return m.group(0) + f"          - {target}\n"
-
-    new_content, count = pattern.subn(add_target, content)
-    return new_content, count
-
-
-def insert_into_shell_loop(target: str, content: str) -> tuple[str, int]:
-    """Insert target into a 'for target in ... ; do' shell loop.
-
-    The last entry in the loop ends with ``; do``.  We change it to end with
-    a backslash continuation and append the new entry with ``; do``.
-
-    Example — before::
-
-              fuzz_block_verification; do
-
-    After::
-
-              fuzz_block_verification \\
-              fuzz_new_target; do
-    """
-    # Match the last fuzz target in the for-loop: "            fuzz_xxx; do"
-    # Indentation: 12 spaces (inside a run: | block).
-    pattern = re.compile(r"(            fuzz_\w+)(; do)", re.MULTILINE)
-
-    # We only want to replace the *last* occurrence (the closing entry).
-    matches = list(pattern.finditer(content))
-    if not matches:
-        return content, 0
-
-    if len(matches) > 1:
-        print(
-            f"  ERROR: found {len(matches)} shell loops matching the pattern; "
-            "cannot determine which one to update. "
-            "Please edit .github/workflows/fuzz.yml manually.",
-            file=sys.stderr,
-        )
-        sys.exit(1)
-
-    m = matches[-1]
-    replacement = f"{m.group(1)} \\\n            {target}{m.group(2)}"
-    new_content = content[: m.start()] + replacement + content[m.end() :]
-    return new_content, 1
-
-
-def insert_into_workflow(target: str, workflow: Path) -> None:
-    """Update all target lists in the fuzz workflow file."""
-    content = workflow.read_text()
-
-    if target in content:
-        print(f"  SKIP .github/workflows/fuzz.yml — {target!r} already present")
-        return
-
-    # 1. YAML matrix blocks (smoke-fuzz, regression)
-    content, yaml_count = insert_into_yaml_matrices(target, content)
-    if yaml_count:
-        print(
-            f"  [+] .github/workflows/fuzz.yml — inserted {target!r} into "
-            f"{yaml_count} YAML matrix block(s)"
-        )
-    else:
-        print(
-            f"  ERROR: no YAML matrix blocks matched in {workflow} — please edit manually",
-            file=sys.stderr,
-        )
-        sys.exit(1)
-
-    # 2. Shell for-loop (perf-baseline)
-    content, loop_count = insert_into_shell_loop(target, content)
-    if loop_count:
-        print(
-            f"  [+] .github/workflows/fuzz.yml — inserted {target!r} into "
-            f"perf-baseline shell loop"
-        )
-    else:
-        print(
-            f"  ERROR: perf-baseline shell loop not found in {workflow} — please edit manually",
-            file=sys.stderr,
-        )
-        sys.exit(1)
-
-    workflow.write_text(content)
-
-
 def main() -> None:
     if len(sys.argv) != 2:
         print(f"Usage: {sys.argv[0]} <TARGET_NAME>", file=sys.stderr)
@@ -167,17 +68,11 @@ def main() -> None:
     root = Path(__file__).parent.parent  # repository root
 
     cargo_toml = root / "fuzz" / "Cargo.toml"
-    workflow = root / ".github" / "workflows" / "fuzz.yml"
-
     if not cargo_toml.exists():
         print(f"ERROR: {cargo_toml} not found", file=sys.stderr)
         sys.exit(1)
-    if not workflow.exists():
-        print(f"ERROR: {workflow} not found", file=sys.stderr)
-        sys.exit(1)
 
     append_cargo_bin(target, cargo_toml)
-    insert_into_workflow(target, workflow)
 
     # ── Print build instructions ──────────────────────────────────────────────
     print()
@@ -197,13 +92,11 @@ def main() -> None:
     print("  4. Run with libFuzzer:  just fuzz-one", target)
     print("     Run with AFL++:      just fuzz-afl", target)
     print()
-    print("  5. This script only edits .github/workflows/fuzz.yml. Add the")
-    print("     target to the other enumeration sites too, then verify with:")
+    print("  5. Every workflow and script derives its target list from")
+    print("     fuzz/Cargo.toml, so no CI edits are needed. Only add a prose row")
+    print("     to the two doc tables, then verify with:")
     print("       python3 scripts/check_target_inventory.py")
     print("     (the same check runs in CI and will fail the build on drift):")
-    print("       .github/workflows/fuzz-afl.yml")
-    print("       .github/workflows/mutants.yml")
-    print("       scripts/mutants-corpus-test.sh")
     print("       README.md, docs/fuzzing.md")
 
 
diff --git a/scripts/check_target_inventory.py b/scripts/check_target_inventory.py
index 7c53ff021..a76d90320 100755
--- a/scripts/check_target_inventory.py
+++ b/scripts/check_target_inventory.py
@@ -1,11 +1,18 @@
 #!/usr/bin/env python3
 """Fail if any fuzz target registered in fuzz/Cargo.toml is missing from a
-workflow / script / doc that enumerates the target list.
+human-authored doc that enumerates the target list.
 
 `fuzz/Cargo.toml` is the single source of truth: every `[[bin]] name = "fuzz_*"`
-must be mentioned by name in each of the consumer files below. This guards
-against the drift that `scripts/add_fuzz_target.py` cannot prevent on its own
-(it only edits `.github/workflows/fuzz.yml`).
+must be mentioned by name in each consumer file below.
+
+The CI workflows and shell scripts derive their target lists *directly* from
+`fuzz/Cargo.toml` at runtime, so they cannot drift and are not checked here:
+  - `.github/workflows/{fuzz,fuzz-afl}.yml` and the `corpus-update.yml` matrices
+    use the `.github/actions/resolve-targets` composite action.
+  - `.github/workflows/mutants.yml` calls the same action for its build loop.
+  - `scripts/mutants-corpus-test.sh` parses `fuzz/Cargo.toml` inline.
+Only the prose target tables in the docs below carry a hand-written description
+per target and therefore need this drift gate.
 
 Usage:
     python3 scripts/check_target_inventory.py
@@ -19,13 +26,10 @@ import re
 import sys
 from pathlib import Path
 
-# Files that enumerate the full target list and must stay in sync with Cargo.toml.
+# Human-authored docs whose prose target tables must stay in sync with Cargo.toml.
+# (Workflows/scripts auto-derive their lists from Cargo.toml — see the module docstring.)
 # Paths are relative to the repository root.
 CONSUMERS = [
-    ".github/workflows/fuzz.yml",
-    ".github/workflows/fuzz-afl.yml",
-    ".github/workflows/mutants.yml",
-    "scripts/mutants-corpus-test.sh",
     "README.md",
     "docs/fuzzing.md",
 ]
diff --git a/scripts/mutants-corpus-test.sh b/scripts/mutants-corpus-test.sh
index 663c8b6fa..0cd962743 100755
--- a/scripts/mutants-corpus-test.sh
+++ b/scripts/mutants-corpus-test.sh
@@ -20,29 +20,23 @@ FUZZ_REPO="${FUZZ_REPO:-"$(cd "${SCRIPT_DIR}/.." && pwd)"}"
 CORPUS_ROOT="${FUZZ_REPO}/corpus/libfuzz"
 FUZZ_DIR="${FUZZ_REPO}/fuzz"
 
-targets=(
-  fuzz_transaction_decoding
-  fuzz_stateless_verification
-  fuzz_state_transition
-  fuzz_block_verification
-  fuzz_encoding_roundtrip
-  fuzz_signature_verification
-  fuzz_replay_prevention
-  fuzz_state_diff_computation
-  fuzz_validate_execute_consistency
-  fuzz_state_serialization
-  fuzz_witness_set_verification
-  fuzz_program_deployment_lifecycle
-  fuzz_apply_state_diff_split_path
-  fuzz_multi_block_state_sequence
-  fuzz_sequencer_vs_replayer
-  fuzz_merkle_tree
-  fuzz_transaction_properties
-  fuzz_privacy_preserving_witness
-  fuzz_encoding_privacy_preserving
-  fuzz_nullifier_set_roundtrip
-  fuzz_privacy_preserving_state_transition
+# Derive the target list from fuzz/Cargo.toml (the single source of truth) — the same
+# `[[bin]] name = "fuzz_*"` parse that .github/actions/resolve-targets uses. This keeps
+# the script in sync with every workflow automatically, with no hand-maintained list.
+# (A while-read loop rather than `mapfile`, so this also works under macOS' bash 3.2.)
+CARGO_TOML="${FUZZ_DIR}/Cargo.toml"
+targets=()
+while IFS= read -r _target; do
+  [ -n "${_target}" ] && targets+=("${_target}")
+done < <(
+  grep -oE 'name = "fuzz_[a-z0-9_]+"' "${CARGO_TOML}" \
+    | sed -E 's/.*"(fuzz_[a-z0-9_]+)"/\1/' \
+    | awk '!seen[$0]++'
 )
+if [ "${#targets[@]}" -eq 0 ]; then
+  echo "ERROR: no fuzz_* [[bin]] targets found in ${CARGO_TOML}" >&2
+  exit 1
+fi
 
 # cargo-fuzz requires the nightly toolchain (-Zsanitizer=address etc.).
 # When this script is called by `cargo-mutants` the working directory is the