diff --git a/.github/scripts/upstream_digest.py b/.github/scripts/upstream_digest.py
new file mode 100644
index 0000000..2f0e949
--- /dev/null
+++ b/.github/scripts/upstream_digest.py
@@ -0,0 +1,158 @@
+#!/usr/bin/env python3
+import json
+import os
+import re
+import subprocess
+import sys
+from datetime import date, timedelta
+
+
+UPSTREAM = "stevearc/oil.nvim"
+UPSTREAM_MD = "doc/upstream.md"
+
+PRS_HEADING = "## Open upstream PRs"
+ISSUES_HEADING = "## Upstream issues"
+
+
+def get_last_tracked_number():
+    try:
+        with open(UPSTREAM_MD) as f:
+            content = f.read()
+        numbers = re.findall(
+            r"\[#(\d+)\]\(https://github\.com/stevearc/oil\.nvim", content
+        )
+        if numbers:
+            return max(int(n) for n in numbers)
+    except OSError:
+        pass
+    return None
+
+
+def gh(*args):
+    result = subprocess.run(
+        ["gh"] + list(args),
+        capture_output=True,
+        text=True,
+        check=True,
+    )
+    return result.stdout.strip()
+
+
+def fetch_items(last_number, since_date):
+    merged_prs = json.loads(
+        gh(
+            "pr", "list",
+            "--repo", UPSTREAM,
+            "--state", "merged",
+            "--limit", "100",
+            "--json", "number,title,url",
+        )
+    )
+    open_prs = json.loads(
+        gh(
+            "pr", "list",
+            "--repo", UPSTREAM,
+            "--state", "open",
+            "--limit", "100",
+            "--json", "number,title,createdAt,url",
+        )
+    )
+    open_issues = json.loads(
+        gh(
+            "issue", "list",
+            "--repo", UPSTREAM,
+            "--state", "open",
+            "--limit", "100",
+            "--json", "number,title,createdAt,url",
+        )
+    )
+
+    if last_number is not None:
+        merged_prs = [x for x in merged_prs if x["number"] > last_number]
+        open_prs = [x for x in open_prs if x["number"] > last_number]
+        open_issues = [x for x in open_issues if x["number"] > last_number]
+    else:
+        cutoff = since_date.isoformat()
+        merged_prs = []
+        open_prs = [x for x in open_prs if x.get("createdAt", "") >= cutoff]
+        open_issues = [x for x in open_issues if x.get("createdAt", "") >= cutoff]
+
+    merged_prs.sort(key=lambda x: x["number"])
+    open_prs.sort(key=lambda x: x["number"])
+    open_issues.sort(key=lambda x: x["number"])
+
+    return merged_prs, open_prs, open_issues
+
+
+def append_to_section(content, heading, new_rows):
+    lines = content.split("\n")
+    in_section = False
+    last_table_row = -1
+
+    for i, line in enumerate(lines):
+        if line.startswith("## "):
+            in_section = line.strip() == heading
+        if in_section and line.startswith("|"):
+            last_table_row = i
+
+    if last_table_row == -1:
+        return content
+
+    return "\n".join(
+        lines[: last_table_row + 1] + new_rows + lines[last_table_row + 1 :]
+    )
+
+
+def main():
+    last_number = get_last_tracked_number()
+    since_date = date.today() - timedelta(days=30)
+
+    merged_prs, open_prs, open_issues = fetch_items(last_number, since_date)
+
+    total = len(merged_prs) + len(open_prs) + len(open_issues)
+    if total == 0:
+        print("No new upstream activity.")
+        return
+
+    with open(UPSTREAM_MD) as f:
+        content = f.read()
+
+    pr_rows = []
+    for pr in open_prs:
+        pr_rows.append(f"| [#{pr['number']}]({pr['url']}) | {pr['title']} | open |")
+    for pr in merged_prs:
+        pr_rows.append(
+            f"| [#{pr['number']}]({pr['url']}) | {pr['title']} | merged — not cherry-picked |"
+        )
+
+    issue_rows = []
+    for issue in open_issues:
+        issue_rows.append(
+            f"| [#{issue['number']}]({issue['url']}) | open | {issue['title']} |"
+        )
+
+    if pr_rows:
+        content = append_to_section(content, PRS_HEADING, pr_rows)
+    if issue_rows:
+        content = append_to_section(content, ISSUES_HEADING, issue_rows)
+
+    with open(UPSTREAM_MD, "w") as f:
+        f.write(content)
+
+    github_output = os.environ.get("GITHUB_OUTPUT")
+    if github_output:
+        with open(github_output, "a") as f:
+            f.write("changed=true\n")
+
+    print(
+        f"Added {len(open_prs)} open PR(s), {len(merged_prs)} merged PR(s), "
+        f"{len(open_issues)} issue(s) to {UPSTREAM_MD}"
+    )
+
+
+if __name__ == "__main__":
+    try:
+        main()
+    except subprocess.CalledProcessError as e:
+        print(f"gh command failed: {e.stderr}", file=sys.stderr)
+        sys.exit(1)
diff --git a/.github/workflows/upstream-digest.yml b/.github/workflows/upstream-digest.yml
new file mode 100644
index 0000000..a70ee5b
--- /dev/null
+++ b/.github/workflows/upstream-digest.yml
@@ -0,0 +1,46 @@
+name: upstream digest
+
+on:
+  schedule:
+    - cron: "0 9 * * 1"
+  workflow_dispatch:
+
+permissions:
+  contents: write
+  pull-requests: write
+
+jobs:
+  digest:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Update upstream tracker
+        id: digest
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: python3 .github/scripts/upstream_digest.py
+
+      - name: Format doc/upstream.md
+        if: steps.digest.outputs.changed == 'true'
+        run: npx --yes prettier --write doc/upstream.md
+
+      - name: Create PR
+        if: steps.digest.outputs.changed == 'true'
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          DATE=$(date +%Y-%m-%d)
+          BRANCH="ci/upstream-digest-${DATE}"
+          git config user.name "github-actions[bot]"
+          git config user.email "github-actions[bot]@users.noreply.github.com"
+          git checkout -b "${BRANCH}"
+          git add doc/upstream.md
+          git commit -m "docs(upstream): add digest for week of ${DATE}"
+          git push origin "${BRANCH}"
+          PR_URL=$(gh pr create \
+            --title "docs(upstream): add digest for week of ${DATE}" \
+            --body "Automated weekly digest of new upstream activity. Triage by updating statuses and notes." \
+            --base main \
+            --head "${BRANCH}")
+          gh pr merge "${PR_URL}" --auto --squash