diff --git a/tests/adr-contract.test.mjs b/tests/adr-contract.test.mjs
new file mode 100644
index 00000000..3a0a2000
--- /dev/null
+++ b/tests/adr-contract.test.mjs
@@ -0,0 +1,217 @@
+import assert from 'node:assert/strict';
+import test from 'node:test';
+import { existsSync, readFileSync, readdirSync } from 'node:fs';
+import { fileURLToPath } from 'node:url';
+import { join } from 'node:path';
+
+const adrDir = fileURLToPath(new URL('../adr', import.meta.url));
+const readme = readFileSync(join(adrDir, 'README.md'), 'utf8');
+
+const ALLOWED_STATUSES = new Set([
+  'Proposed',
+  'Accepted',
+  'Rejected',
+  'Superseded',
+  'Deprecated',
+]);
+
+const markdownFiles = readdirSync(adrDir)
+  .filter((name) => name.endsWith('.md') && name !== 'README.md')
+  .sort();
+
+// Files whose name begins with a four-digit sequence number claim an ADR
+// number; supporting material that does not is exempt from the number and
+// index contracts.
+const numberPrefixed = markdownFiles.filter((name) => /^\d{4}-/.test(name));
+
+// | [0001](./0001-slug.md) | Title | Status |
+const INDEX_ROW =
+  /^\|\s*\[(\d{4})\]\((\.\/[^)]+)\)\s*\|\s*(.+?)\s*\|\s*(\S+)\s*\|\s*$/;
+
+const indexRows = readme
+  .split('\n')
+  .map((line) => line.match(INDEX_ROW))
+  .filter(Boolean)
+  .map(([, number, href, title, status]) => ({
+    number,
+    href,
+    title,
+    status,
+  }));
+
+function readAdr(name) {
+  return readFileSync(join(adrDir, name), 'utf8');
+}
+
+function headerField(body, field) {
+  const match = body.match(
+    new RegExp(`^- \\*\\*${field}\\*\\*:\\s*(.+)$`, 'm'),
+  );
+  return match ? match[1].trim() : null;
+}
+
+function heading(body) {
+  const match = body.match(/^#\s+(.+)$/m);
+  return match ? match[1].trim() : null;
+}
+
+test('the index table lists at least one ADR', () => {
+  assert.ok(
+    indexRows.length > 0,
+    'adr/README.md must contain a table of ADR rows matching | [NNNN](./file.md) | Title | Status |',
+  );
+});
+
+test('every indexed ADR links to a file that exists', () => {
+  for (const row of indexRows) {
+    const target = row.href.replace(/^\.\//, '');
+    assert.ok(
+      existsSync(join(adrDir, target)),
+      `adr/README.md row ${row.number} links to missing file ${row.href}`,
+    );
+  }
+});
+
+test('every indexed ADR links to a file named for its number', () => {
+  for (const row of indexRows) {
+    const target = row.href.replace(/^\.\//, '');
+    assert.ok(
+      target.startsWith(`${row.number}-`),
+      `adr/README.md row ${row.number} links to ${target}, which is not named for ADR ${row.number}`,
+    );
+  }
+});
+
+test('index numbers are unique', () => {
+  const seen = new Set();
+  for (const row of indexRows) {
+    assert.ok(
+      !seen.has(row.number),
+      `adr/README.md lists ADR ${row.number} more than once`,
+    );
+    seen.add(row.number);
+  }
+});
+
+// A decision record is a number-prefixed file that either heads itself as an
+// ADR or is linked from the index; supporting drafts filed alongside them are
+// not held to the ADR header contract.
+const adrFiles = numberPrefixed.filter(
+  (name) =>
+    /^ADR\s+\d{4}:/.test(heading(readAdr(name)) ?? '') ||
+    indexRows.some((row) => row.href.replace(/^\.\//, '') === name),
+);
+
+test('at least one decision record is present', () => {
+  assert.ok(adrFiles.length > 0, 'adr/ contains no ADR files');
+});
+
+test('every ADR file carries a parseable Date header', () => {
+  for (const name of adrFiles) {
+    const date = headerField(readAdr(name), 'Date');
+    assert.ok(date, `adr/${name} is missing a "- **Date**:" header field`);
+    assert.ok(
+      !Number.isNaN(Date.parse(date)),
+      `adr/${name} has an unparseable Date header: ${date}`,
+    );
+  }
+});
+
+test('every ADR file declares an allowed Status', () => {
+  for (const name of adrFiles) {
+    const status = headerField(readAdr(name), 'Status');
+    assert.ok(status, `adr/${name} is missing a "- **Status**:" header field`);
+    assert.ok(
+      ALLOWED_STATUSES.has(status),
+      `adr/${name} has status "${status}"; expected one of ${[...ALLOWED_STATUSES].join(', ')}`,
+    );
+  }
+});
+
+test('index status matches the status recorded in each ADR file', () => {
+  for (const row of indexRows) {
+    const target = row.href.replace(/^\.\//, '');
+    if (!existsSync(join(adrDir, target))) continue;
+    const status = headerField(readAdr(target), 'Status');
+    assert.equal(
+      row.status,
+      status,
+      `adr/README.md lists ADR ${row.number} as "${row.status}" but adr/${target} records "${status}"`,
+    );
+  }
+});
+
+test('index title matches the ADR heading', () => {
+  for (const row of indexRows) {
+    const target = row.href.replace(/^\.\//, '');
+    if (!existsSync(join(adrDir, target))) continue;
+    const title = heading(readAdr(target));
+    assert.ok(title, `adr/${target} is missing an H1 heading`);
+    const expected = title.replace(/^ADR\s+\d{4}:\s*/, '');
+    assert.equal(
+      row.title,
+      expected,
+      `adr/README.md titles ADR ${row.number} "${row.title}" but adr/${target} heads "${expected}"`,
+    );
+  }
+});
+
+test('every ADR heading names its own file number', () => {
+  for (const name of adrFiles) {
+    const title = heading(readAdr(name));
+    const declared = title?.match(/^ADR\s+(\d{4}):/)?.[1];
+    if (!declared) continue;
+    assert.equal(
+      declared,
+      name.slice(0, 4),
+      `adr/${name} heads itself "ADR ${declared}" but is filed under ${name.slice(0, 4)}`,
+    );
+  }
+});
+
+test('relative links between ADR files resolve', () => {
+  for (const name of [...markdownFiles, 'README.md']) {
+    const body = readFileSync(join(adrDir, name), 'utf8');
+    for (const [, href] of body.matchAll(/]\((\.\/[^)#]+)[^)]*\)/g)) {
+      const target = href.replace(/^\.\//, '');
+      assert.ok(
+        existsSync(join(adrDir, target)),
+        `adr/${name} links to ${href}, which does not exist`,
+      );
+    }
+  }
+});
+
+// Known deviation, tracked separately: adr/0001-stakeholder-outreach-draft.md
+// takes the 0001 prefix without being ADR 0001, so the directory holds two
+// files claiming that number and one of them is absent from the index.
+test('each ADR number is claimed by exactly one file', { todo: true }, () => {
+  const byNumber = new Map();
+  for (const name of numberPrefixed) {
+    const number = name.slice(0, 4);
+    byNumber.set(number, [...(byNumber.get(number) ?? []), name]);
+  }
+  for (const [number, names] of byNumber) {
+    assert.equal(
+      names.length,
+      1,
+      `ADR number ${number} is claimed by ${names.length} files: ${names.join(', ')}`,
+    );
+  }
+});
+
+test(
+  'every number-prefixed ADR file is listed in the index',
+  { todo: true },
+  () => {
+    const indexed = new Set(
+      indexRows.map((row) => row.href.replace(/^\.\//, '')),
+    );
+    for (const name of numberPrefixed) {
+      assert.ok(
+        indexed.has(name),
+        `adr/${name} is not listed in adr/README.md`,
+      );
+    }
+  },
+);