fix(scripts): select sandbox containers by anchored name (#7189)

README.md

@@ -770,6 +770,10 @@ DeerFlow runs the agent runtime inside the Gateway API. Development mode enables
 | **Stop** | `./scripts/serve.sh --stop`<br/>`make stop` | `./scripts/docker.sh stop`<br/>`make docker-stop` | `./scripts/deploy.sh down`<br/>`make down` |
 | **Restart** | `./scripts/serve.sh --restart [flags]` | `./scripts/docker.sh restart` | — |
 
+Local and Docker development stop commands also stop running sandbox containers
+whose names start with `deer-flow-sandbox-`. Containers with that text elsewhere
+in their names are left alone.
+
 `make start` and `make start-daemon` rebuild the frontend with `next build` on
 every run. To reuse the last build instead, pass `SKIP_FRONTEND_BUILD=1` (or add
 `--skip-frontend-build` when calling `./scripts/serve.sh --prod` directly). This

backend/tests/test_cleanup_containers_script.py

@@ -0,0 +1,62 @@
+"""Offline checks for the names selected by the sandbox cleanup helper."""
+
+from __future__ import annotations
+
+import json
+import os
+import subprocess
+from pathlib import Path
+
+import pytest
+
+SCRIPT = Path(__file__).resolve().parents[2] / "scripts" / "cleanup-containers.sh"
+
+
+def _stub(path: Path, name: str, source: str) -> None:
+    executable = path / name
+    executable.write_text("#!/bin/sh\n" + source, encoding="utf-8", newline="\n")
+    executable.chmod(0o755)
+
+
+@pytest.mark.skipif(os.name == "nt", reason="POSIX executable stubs")
+def test_docker_cleanup_stops_only_sandbox_prefix(tmp_path: Path) -> None:
+    bin_dir = tmp_path / "bin"
+    bin_dir.mkdir()
+    stop_log = tmp_path / "stopped"
+    _stub(
+        bin_dir,
+        "docker",
+        'if [ "$1" = ps ]; then\n'
+        '  printf "other-id not-deer-flow-sandbox Up 3 days\\nvalid-id deer-flow-sandbox-abc Up 1 hour\\nbackup-id deer-flow-sandboxbackup Up 2 days\\n"\n'
+        'elif [ "$1" = stop ]; then\n'
+        '  shift; printf "%s\\n" "$@" >> "$STOP_LOG"\n'
+        "fi\n",
+    )
+    env = {**os.environ, "PATH": f"{bin_dir}:/usr/bin:/bin", "STOP_LOG": str(stop_log)}
+    result = subprocess.run(["bash", str(SCRIPT), "deer-flow-sandbox"], env=env, capture_output=True, text=True, timeout=5)
+
+    assert result.returncode == 0, result.stderr
+    assert stop_log.read_text(encoding="utf-8").splitlines() == ["valid-id"]
+    assert "not-deer-flow-sandbox" not in result.stdout
+    assert "deer-flow-sandbox-abc Up 1 hour" in result.stdout
+
+
+@pytest.mark.skipif(os.name == "nt", reason="POSIX executable stubs")
+def test_apple_cleanup_ignores_unrelated_names_and_bad_entries(tmp_path: Path) -> None:
+    bin_dir = tmp_path / "bin"
+    bin_dir.mkdir()
+    stop_log = tmp_path / "stopped"
+    entries = [
+        {"configuration": None},
+        {"configuration": {"id": "not-deer-flow-sandbox"}},
+        {"configuration": {"id": "deer-flow-sandbox-abc"}},
+    ]
+    payload = tmp_path / "containers.json"
+    payload.write_text(json.dumps(entries), encoding="utf-8")
+    _stub(bin_dir, "container", 'if [ "$1" = list ]; then cat "$CONTAINERS_JSON"; elif [ "$1" = stop ]; then printf "%s\\n" "$2" >> "$STOP_LOG"; fi\n')
+    env = {**os.environ, "PATH": f"{bin_dir}:/usr/bin:/bin", "STOP_LOG": str(stop_log), "CONTAINERS_JSON": str(payload)}
+    result = subprocess.run(["bash", str(SCRIPT), "deer-flow-sandbox"], env=env, capture_output=True, text=True, timeout=5)
+
+    assert result.returncode == 0, result.stderr
+    assert stop_log.read_text(encoding="utf-8").splitlines() == ["deer-flow-sandbox-abc"]
+    assert "not-deer-flow-sandbox" not in result.stdout

scripts/AGENTS.md

@@ -214,6 +214,11 @@ runs `check-data-not-in-use.sh` before `make stop` (which would stop a live
 stack's sandboxes) and refuses while that container runs; an absent or
 unreachable Docker passes. `make stop` must still run before the delete: it
 stops `deer-flow-sandbox*` containers, whose thread mounts live in that tree.
+`cleanup-containers.sh` must verify an anchored `<prefix>-` name after Docker's
+substring `name=` filter, and use the same rule for Apple Container IDs. Parse
+Apple Container JSON without interpolating the prefix into Python source; bad
+entries must not hide later valid containers. Offline stub coverage lives in
+`backend/tests/test_cleanup_containers_script.py`.
 `RUNTIME_DATA_CONTENTS` feeds both the help line and the deletion notice.
 
 Host-side pnpm calls must go through `scripts/pnpm.py`. With native Windows

scripts/cleanup-containers.sh

@@ -22,14 +22,23 @@ echo "Cleaning up sandbox containers with prefix: ${PREFIX}"
 cleanup_docker() {
     if command -v docker &> /dev/null; then
         echo -n "Checking Docker containers... "
-        DOCKER_CONTAINERS=$(docker ps -q --filter "name=${PREFIX}" 2>/dev/null || echo "")
+        # Docker's name filter matches substrings. Check the returned names as
+        # well so a container like not-deer-flow-sandbox is never stopped.
+        DOCKER_LIST=$(docker ps --filter "name=${PREFIX}-" --format '{{.ID}} {{.Names}} {{.Status}}' 2>/dev/null || true)
+        docker_ids=()
+        docker_rows=()
+        while read -r cid name status; do
+            case "$name" in
+                "$PREFIX"-*) docker_ids+=("$cid"); docker_rows+=("$cid $name $status") ;;
+            esac
+        done <<< "$DOCKER_LIST"
 
-        if [ -n "$DOCKER_CONTAINERS" ]; then
+        if [ "${#docker_ids[@]}" -gt 0 ]; then
             echo ""
             echo "Found Docker containers to clean up:"
-            docker ps --filter "name=${PREFIX}" --format "table {{.ID}}\t{{.Names}}\t{{.Status}}"
+            printf '  %s\n' "${docker_rows[@]}"
             echo "Stopping Docker containers..."
-            echo "$DOCKER_CONTAINERS" | xargs docker stop 2>/dev/null || true
+            docker stop "${docker_ids[@]}" >/dev/null 2>&1 || true
             echo -e "${GREEN}✓ Docker containers stopped${NC}"
         else
             echo -e "${GREEN}none found${NC}"
@@ -49,21 +58,23 @@ cleanup_apple_container() {
 
         if [ "$CONTAINER_LIST" != "[]" ] && [ -n "$CONTAINER_LIST" ]; then
             # Extract container IDs that match our prefix
-            CONTAINER_IDS=$(echo "$CONTAINER_LIST" | python3 -c "
+            CONTAINER_IDS=$(echo "$CONTAINER_LIST" | CLEANUP_CONTAINER_PREFIX="$PREFIX" python3 -c '
 import json
+import os
 import sys
 try:
     containers = json.load(sys.stdin)
     if isinstance(containers, list):
+        prefix = os.environ["CLEANUP_CONTAINER_PREFIX"] + "-"
         for c in containers:
             if isinstance(c, dict):
-                # Apple Container uses 'id' field which contains the container name
-                cid = c.get('configuration').get('id', '')
-                if '${PREFIX}' in cid:
+                configuration = c.get("configuration")
+                cid = configuration.get("id") if isinstance(configuration, dict) else None
+                if isinstance(cid, str) and cid.startswith(prefix):
                     print(cid)
 except:
     pass
-" 2>/dev/null || echo "")
+' 2>/dev/null || echo "")
 
             if [ -n "$CONTAINER_IDS" ]; then
                 echo ""