diff --git a/README.md b/README.md
index 7990fb5..43230d7 100644
--- a/README.md
+++ b/README.md
@@ -88,7 +88,7 @@ print(json.dumps(ComputerState.model_json_schema(), indent=2))
 ```
 
 The same schemas ship as JSON under `openadapt_types/schemas/` for TypeScript,
-Rust, and anything else that isn't Python. Nineteen files, including
+Rust, and anything else that isn't Python. Seventeen files, including
 `execute-v1-openapi.json`, the public OpenAdapt Execute contract.
 
 ## Converting from the older formats
diff --git a/tests/test_readme_claims.py b/tests/test_readme_claims.py
new file mode 100644
index 0000000..427bba0
--- /dev/null
+++ b/tests/test_readme_claims.py
@@ -0,0 +1,66 @@
+"""The README states facts about the package. Keep them true.
+
+A count written by hand goes stale the moment a schema is added or removed,
+and a reader has no way to tell. These tests read the claim out of the README
+and check it against the shipped tree.
+"""
+
+from __future__ import annotations
+
+import re
+from pathlib import Path
+
+ROOT = Path(__file__).resolve().parents[1]
+README = ROOT / "README.md"
+SCHEMA_DIR = ROOT / "openadapt_types" / "schemas"
+
+# Indexed by value, so WORDS[17] == "seventeen". Extend it when the directory
+# grows past the end rather than dropping the count from the README.
+WORDS = (
+    "zero", "one", "two", "three", "four", "five", "six", "seven", "eight",
+    "nine", "ten", "eleven", "twelve", "thirteen", "fourteen", "fifteen",
+    "sixteen", "seventeen", "eighteen", "nineteen", "twenty",
+    "twenty-one", "twenty-two", "twenty-three", "twenty-four", "twenty-five",
+    "twenty-six", "twenty-seven", "twenty-eight", "twenty-nine", "thirty",
+)
+
+# Tied to the sentence itself, so an unrelated "N files" elsewhere in the
+# README cannot satisfy this test by accident.
+SCHEMA_COUNT_CLAIM = re.compile(
+    r"([A-Za-z][A-Za-z-]*) files, including\s+`execute-v1-openapi\.json`"
+)
+
+
+def test_readme_schema_file_count_matches_the_shipped_directory() -> None:
+    actual = len(list(SCHEMA_DIR.glob("*.json")))
+    assert actual < len(WORDS), (
+        f"{SCHEMA_DIR.name}/ holds {actual} JSON files, past the end of WORDS. "
+        "Add the number word and update the README."
+    )
+
+    match = SCHEMA_COUNT_CLAIM.search(README.read_text(encoding="utf-8"))
+    assert match is not None, (
+        "The README no longer states how many JSON files ship under "
+        f"{SCHEMA_DIR.name}/. Restore the claim or delete this test."
+    )
+
+    claimed = match.group(1).lower()
+    assert claimed == WORDS[actual], (
+        f"README says {claimed!r} JSON files ship under {SCHEMA_DIR.name}/, "
+        f"but {actual} are there. Write {WORDS[actual]!r}."
+    )
+
+
+def test_readme_only_links_to_files_that_exist() -> None:
+    text = README.read_text(encoding="utf-8")
+    targets = re.findall(r"\]\((?!https?:)([^)#]+)", text)
+    missing = [t for t in targets if not (ROOT / t).exists()]
+    assert not missing, f"README links to paths that do not exist: {missing}"
+
+
+def test_readme_names_only_schemas_that_ship() -> None:
+    text = README.read_text(encoding="utf-8")
+    named = set(re.findall(r"`([a-z0-9-]+\.json)`", text))
+    shipped = {path.name for path in SCHEMA_DIR.glob("*.json")}
+    missing = sorted(named - shipped)
+    assert not missing, f"README names schema files that do not ship: {missing}"