diff --git a/.github/workflows/typos.yml b/.github/workflows/typos.yml
new file mode 100644
index 000000000..a26d2df3f
--- /dev/null
+++ b/.github/workflows/typos.yml
@@ -0,0 +1,20 @@
+name: Spell check (typos)
+
+on:
+  workflow_dispatch:
+  push:
+    branches:
+      - "main"
+  pull_request:
+
+permissions:
+  contents: read
+
+jobs:
+  typos:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@08c6903cd8c0fde910a37f88322edcfb5dd907a8 # pin@v4
+
+      - name: Run typos spell checker
+        uses: crate-ci/typos@631208b7aac2daa8b707f55e7331f9112b0e062d # pin@v1.44.0
diff --git a/.typos.toml b/.typos.toml
new file mode 100644
index 000000000..dbd4bf271
--- /dev/null
+++ b/.typos.toml
@@ -0,0 +1,32 @@
+# Configuration for the typos spell checker.
+# See https://github.com/crate-ci/typos
+
+[files]
+# Exclude generated, binary, and data files that produce false positives.
+extend-exclude = [
+    "*.json",
+    "*.lock",
+    "tests_data/",
+    "assets/models/",
+]
+
+[default.extend-words]
+# Domain-specific terms or intentional spellings kept to avoid false positives.
+
+# "hve" appears in ONNX model internals referenced in comments.
+hve = "hve"
+
+# "afterall" is used as a variable/identifier name in test scripts.
+afterall = "afterall"
+
+# "Hashi" is the first half of "HashiCorp" (vendor name in content type descriptions).
+hashi = "hashi"
+
+# "Harc" is the first half of "LHarc" (a legacy archive format name).
+harc = "harc"
+
+# "cpy" / "CPY" are standard file extensions for COBOL copybook files.
+cpy = "cpy"
+
+# "LICENSEs" is the plural of "LICENSE" (used in dist-workspace.toml comments).
+licens = "licens"
diff --git a/python/tests/test_magika_python_module.py b/python/tests/test_magika_python_module.py
index e25bd40b6..7356b3663 100644
--- a/python/tests/test_magika_python_module.py
+++ b/python/tests/test_magika_python_module.py
@@ -274,7 +274,6 @@ def test_magika_module_with_whitespaces() -> None:
     )
 
     for ws_num in ws_nums:
-        print(f"Calling indentify_bytes with {ws_num} whitespaces")
         content = b" " * ws_num
         res = m.identify_bytes(content)
         assert (