diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index e7caaee..59d2059 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -124,11 +124,11 @@ jobs:
     - name: Test CRDT Synthesis (fixed)
       shell: nix develop --command bash -e {0}
       run: |
-        python -m tests.synthesize_crdt synth g_set fixed
-        python -m tests.synthesize_crdt synth 2p_set fixed
-        python -m tests.synthesize_crdt synth flag_ew fixed
-        python -m tests.synthesize_crdt synth flag_dw fixed
-        python -m tests.synthesize_crdt synth grow_only_counter fixed
+        python -m tests.synthesize_crdt synth g_set --fixed
+        python -m tests.synthesize_crdt synth 2p_set --fixed
+        python -m tests.synthesize_crdt synth flag_ew --fixed
+        python -m tests.synthesize_crdt synth flag_dw --fixed
+        python -m tests.synthesize_crdt synth grow_only_counter --fixed
 
     - name: Save Nix cache
       if: steps.cache.outputs.cache-hit != 'true'
diff --git a/README.md b/README.md
index 152e318..6f5f991 100644
--- a/README.md
+++ b/README.md
@@ -61,3 +61,14 @@ For example, we can synthesize for the 2P-Set benchmark with
 ```bash
 python -m tests.synthesize_crdt synth 2p_set
 ```
+
+In general, you can use the following command structure:
+```bash
+python -m tests.synthesize_crdt <mode> <benchmark> [--fixed] [--first <N>] [--repeat <N>]
+```
+Where:
+- `<mode>` is either `synth` for bounded synthesis with pruning or `synth-unbounded` for direct unbounded synthesis.
+- `<benchmark>` is the name of the benchmark or `all` to run all benchmarks.
+- `--fixed` (optional) uses a fixed lattice structure instead of exploring all structures.
+- `--first <N>` (optional) synthesizes the first N structures.
+- `--repeat <N>` (optional) specifies the number of repetitions for the synthesis process.
diff --git a/tests/synthesize_crdt.py b/tests/synthesize_crdt.py
index a0cab8d..d1f89a3 100644
--- a/tests/synthesize_crdt.py
+++ b/tests/synthesize_crdt.py
@@ -1,3 +1,4 @@
+import argparse
 import contextlib
 import csv
 from time import time
@@ -261,12 +262,21 @@ def increasing_depth_structures(underlying):
             yield (base_depth, struct)
         base_depth += 1
 
-if __name__ == "__main__":
-    mode = sys.argv[1]
-    bench = sys.argv[2]
-    fixed_structure = len(sys.argv) >= 4 and sys.argv[3] == "fixed"
-    first_n = int(sys.argv[3].split("first_")[1]) if (len(sys.argv) >= 4 and sys.argv[3].startswith("first_")) else None
-    reps = int(sys.argv[3].split("repeat_")[1]) if (len(sys.argv) >= 4 and sys.argv[3].startswith("repeat_")) else 1
+def main():
+    parser = argparse.ArgumentParser(description='Synthesize CRDTs from sequential types.')
+    parser.add_argument('mode', choices=['synth', 'synth-unbounded'], help='synthesis mode')
+    parser.add_argument('benchmark', help='benchmark name or "all"')
+    parser.add_argument('--fixed', action='store_true', help='use fixed lattice structure')
+    parser.add_argument('--first', type=int, help='synthesize the first N structures')
+    parser.add_argument('--repeat', type=int, default=1, help='number of repetitions')
+
+    args = parser.parse_args()
+
+    mode = args.mode
+    bench = args.benchmark
+    fixed_structure = args.fixed
+    first_n = args.first
+    reps = args.repeat
 
     if bench == "all":
         benches = list(benchmarks.keys())
@@ -353,3 +363,6 @@ def increasing_depth_structures(underlying):
                             for (i, time) in enumerate(times):
                                 percent = (i + 1) / len(times)
                                 distribution_file.write(f"{time},{percent}\n")
+
+if __name__ == "__main__":
+    main()