fix: address review findings on fetch-url-tool

- fetch_url: revalidate allowlist on every redirect hop (fixes SSRF where
  302 to disallowed host / 169.254.169.254 / file:// bypassed the
  url_allowlist). Stream response body with iter_bytes and cap at 1MB
  so a multi-GB response from an allowlisted host cannot OOM the service.
  Reuses the manual-redirect-loop pattern from yarn_client.

- list_applications: stop swallowing 404 (YARN returns 200+empty for
  "no match"; 404 means the RM doesn't support the endpoint — surface
  the YarnError instead of hiding it as an empty result). Add
  Field(ge=1, le=10000) to ListApplicationsRequest.limit so a runaway
  limit is rejected at the Pydantic layer with 422.

- save_connection: PATCH semantics for existing records. Re-route to
  update_connection when the name already exists so partial updates
  (e.g. only master) no longer wipe url_allowlist back to []. Uses an
  _UNSET sentinel in the tool function to distinguish "omitted" from
  "None" without breaking the existing parameter list.

- README: drop leading space on 5 new connection-tool table rows that
  was breaking GitHub Flavored Markdown table continuity.

- Indentation: normalize connections.py and requests.py to 4-space
  indent (auth_password/auth_principal/auth_keytab were 3-space).

Co-Authored-By: Claude <noreply@anthropic.com>

README.md

@@ -100,11 +100,11 @@ MCP 客户端需要先执行 `initialize` 握手，拿到 `mcp-session-id` 后
 
 | 工具 | 作用 |
 | --- | --- |
- | `save_connection` | 保存或更新一个 Spark/YARN 连接配置 |
- | `list_connections` | 列出所有连接配置 |
- | `get_connection` | 按名称读取连接配置 |
- | `delete_connection` | 删除连接配置 |
- | `update_connection` | 部分更新一个已存在的连接 (PATCH 语义, 只改提供的字段) |
+| `save_connection` | 保存或更新一个 Spark/YARN 连接配置 |
+| `list_connections` | 列出所有连接配置 |
+| `get_connection` | 按名称读取连接配置 |
+| `delete_connection` | 删除连接配置 |
+| `update_connection` | 部分更新一个已存在的连接 (PATCH 语义, 只改提供的字段) |
 | `write_job_file` | 将 LLM 已生成的 PySpark 代码写入服务端文件 |
 | `read_job_file` | 读取已存在的 PySpark 脚本内容 |
 | `update_job_file` | 覆盖更新已存在的 PySpark 脚本 |

spark_executor/core/yarn_client.py

@@ -299,9 +299,6 @@ def list_applications(
     resp = _request("GET", url, params=params,
                     verify=config.verify_for_httpx(),
                     auth=config.auth_for_httpx())
-    if resp.status_code == 404:
-        # No apps match (or RM doesn't support the endpoint)
-        return []
     if resp.status_code >= 400:
         raise YarnError(
             f"YARN list applications failed: {resp.status_code} {resp.text[:200]}"

spark_executor/server.py

@@ -390,12 +390,20 @@ def _list_applications(req: ListApplicationsRequest):
         "yarn_rm_url is required for get_external_* to work** — spark-submit "
         "can discover the RM for submissions, but direct YARN REST queries "
         "need an explicit URL. Saving with an existing name overwrites the "
-        "record in place (no version history)."
+        "record in place (no version history). When the name already exists, "
+        "only the provided fields are changed (PATCH semantics); omitted "
+        "fields keep their previous values."
     ),
 )
 def _save_connection(req: SaveConnectionRequest):
+    fields = req.model_dump(exclude_none=True)
+    name = fields.pop("name")
+    try:
+        get_connection(name)
+    except KeyError:
         # exclude_none so we don't overwrite the function's default with explicit None
-    return save_connection(**req.model_dump(exclude_none=True))
+        return save_connection(name=name, **fields)
+    return update_connection(name=name, **fields)
 
 
 @app.post(

spark_executor/tools/connections.py

@@ -4,45 +4,75 @@
 @Author :tao.chen
 """
 from common.logging import logger
-from spark_executor.core.connection_store import ConnectionStore, store
+from spark_executor.core.connection_store import store
 from spark_executor.models import Connection
 
 
+_UNSET = object()
+
+
 def save_connection(
     *,
     name: str,
     master: str,
-    deploy_mode: str = "cluster",
-    yarn_rm_url: str | None = None,
-    spark_conf: dict[str, str] | None = None,
-    ssl_verify: bool | None = None,
-    ssl_ca_bundle: str | None = None,
-    auth_type: str = "none",
-    auth_user: str | None = None,
-   auth_password: str | None = None,
-   auth_principal: str | None = None,
-   auth_keytab: str | None = None,
-    url_allowlist: list[str] | None = None,
-) -> dict[str, str]:
+    deploy_mode: str = _UNSET,  # type: ignore[assignment]
+    yarn_rm_url: str | None = _UNSET,  # type: ignore[assignment]
+    spark_conf: dict[str, str] | None = _UNSET,  # type: ignore[assignment]
+    ssl_verify: bool | None = _UNSET,  # type: ignore[assignment]
+    ssl_ca_bundle: str | None = _UNSET,  # type: ignore[assignment]
+    auth_type: str = _UNSET,  # type: ignore[assignment]
+    auth_user: str | None = _UNSET,  # type: ignore[assignment]
+    auth_password: str | None = _UNSET,  # type: ignore[assignment]
+    auth_principal: str | None = _UNSET,  # type: ignore[assignment]
+    auth_keytab: str | None = _UNSET,  # type: ignore[assignment]
+    url_allowlist: list[str] | None = _UNSET,  # type: ignore[assignment]
+) -> dict[str, object]:
     logger.debug(
-        f"save_connection enter name={name} master={master} deploy_mode={deploy_mode} "
-        f"yarn_rm_url={yarn_rm_url} spark_conf_keys={list((spark_conf or {}).keys())}"
-    )
-    conn = Connection(
-        name=name,
-        master=master,
-        deploy_mode=deploy_mode,
-        yarn_rm_url=yarn_rm_url,
-        spark_conf=spark_conf or {},
-        ssl_verify=ssl_verify,
-        ssl_ca_bundle=ssl_ca_bundle,
-        auth_type=auth_type,
-        auth_user=auth_user,
-        auth_password=auth_password,
-       auth_principal=auth_principal,
-       auth_keytab=auth_keytab,
-        url_allowlist=url_allowlist or [],
+        f"save_connection enter name={name} master={master} "
+        f"spark_conf_keys={list((spark_conf if isinstance(spark_conf, dict) else {}).keys())}"
     )
+    existing = store.get(name)
+    if existing is not None:
+        fields: dict[str, object] = {"master": master}
+        if deploy_mode is not _UNSET:
+            fields["deploy_mode"] = deploy_mode
+        if yarn_rm_url is not _UNSET:
+            fields["yarn_rm_url"] = yarn_rm_url
+        if spark_conf is not _UNSET:
+            fields["spark_conf"] = spark_conf or {}
+        if ssl_verify is not _UNSET:
+            fields["ssl_verify"] = ssl_verify
+        if ssl_ca_bundle is not _UNSET:
+            fields["ssl_ca_bundle"] = ssl_ca_bundle
+        if auth_type is not _UNSET:
+            fields["auth_type"] = auth_type
+        if auth_user is not _UNSET:
+            fields["auth_user"] = auth_user
+        if auth_password is not _UNSET:
+            fields["auth_password"] = auth_password
+        if auth_principal is not _UNSET:
+            fields["auth_principal"] = auth_principal
+        if auth_keytab is not _UNSET:
+            fields["auth_keytab"] = auth_keytab
+        if url_allowlist is not _UNSET:
+            fields["url_allowlist"] = url_allowlist or []
+        return update_connection(name, **fields)
+    new_fields: dict[str, object] = {
+        "name": name,
+        "master": master,
+        "deploy_mode": deploy_mode if deploy_mode is not _UNSET else "cluster",
+        "yarn_rm_url": yarn_rm_url if yarn_rm_url is not _UNSET else None,
+        "spark_conf": (spark_conf if spark_conf is not _UNSET else None) or {},
+        "ssl_verify": ssl_verify if ssl_verify is not _UNSET else None,
+        "ssl_ca_bundle": ssl_ca_bundle if ssl_ca_bundle is not _UNSET else None,
+        "auth_type": auth_type if auth_type is not _UNSET else "none",
+        "auth_user": auth_user if auth_user is not _UNSET else None,
+        "auth_password": auth_password if auth_password is not _UNSET else None,
+        "auth_principal": auth_principal if auth_principal is not _UNSET else None,
+        "auth_keytab": auth_keytab if auth_keytab is not _UNSET else None,
+        "url_allowlist": (url_allowlist if url_allowlist is not _UNSET else None) or [],
+    }
+    conn = Connection(**new_fields)
     store.save(conn)
     return {"name": name, "status": "SAVED"}
 

spark_executor/tools/fetch_url.py

@@ -12,7 +12,7 @@ are intentionally NOT performed. Reuses the connection's saved auth/SSL
 config so the agent doesn't need cluster credentials.
 """
 import fnmatch
-from urllib.parse import urlparse
+from urllib.parse import urlparse, urljoin
 
 import httpx
 
@@ -24,6 +24,9 @@ from spark_executor.tools.connections import store as conn_store
 _MAX_BODY_BYTES = 1_000_000  # 1 MB cap on response body
 _REQUEST_TIMEOUT_SECONDS = 30
 
+_MAX_REDIRECTS = 3
+_REDIRECT_STATUSES = frozenset({301, 302, 303, 307, 308})
+
 
 def _host_matches_any_glob(host: str, allowlist: list[str]) -> bool:
     """True if `host` matches any of the fnmatch glob patterns.
@@ -67,6 +70,29 @@ def _validate_url_host(url: str, allowlist: list[str] | None) -> None:
     )
 
 
+def _read_streamed_body(resp: httpx.Response) -> tuple[bytes, bool]:
+    """Stream the response body, keeping at most ``_MAX_BODY_BYTES`` bytes.
+
+    Returns ``(body_bytes, truncated)``. Stops reading as soon as the cap is
+    exceeded so that a multi-gigabyte response from an allowlisted host cannot
+    OOM the service.
+    """
+    chunks: list[bytes] = []
+    total = 0
+    truncated = False
+    for chunk in resp.iter_bytes():
+        if total + len(chunk) <= _MAX_BODY_BYTES:
+            chunks.append(chunk)
+            total += len(chunk)
+        else:
+            if total < _MAX_BODY_BYTES:
+                chunks.append(chunk[: _MAX_BODY_BYTES - total])
+                total = _MAX_BODY_BYTES
+            truncated = True
+            break
+    return b"".join(chunks), truncated
+
+
 def fetch_url(url: str, connection_name: str) -> FetchUrlResult:
     """Proxy an HTTP GET to url using the auth/SSL settings of connection_name."""
     logger.debug(f"fetch_url enter url={url} connection_name={connection_name}")
@@ -77,16 +103,22 @@ def fetch_url(url: str, connection_name: str) -> FetchUrlResult:
     _validate_url_host(url, conn.url_allowlist)
 
     config = YarnClientConfig.from_connection(conn)
-    resp = httpx.get(
-        url,
-        auth=config.auth_for_httpx(),
-        verify=config.verify_for_httpx(),
-        timeout=_REQUEST_TIMEOUT_SECONDS,
-        follow_redirects=True,
-    )
+    auth = config.auth_for_httpx()
+    verify = config.verify_for_httpx()
 
-    body = resp.text[:_MAX_BODY_BYTES]
-    truncated = len(resp.text) > _MAX_BODY_BYTES
+    for hop in range(_MAX_REDIRECTS + 1):
+        with httpx.stream(
+            "GET",
+            url,
+            auth=auth,
+            verify=verify,
+            timeout=_REQUEST_TIMEOUT_SECONDS,
+            follow_redirects=False,
+        ) as resp:
+            if resp.status_code not in _REDIRECT_STATUSES:
+                body_bytes, truncated = _read_streamed_body(resp)
+                encoding = resp.encoding or "utf-8"
+                body = body_bytes.decode(encoding, errors="replace")
 
                 logger.info(
                     f"fetch_url ok url={url} connection_name={connection_name} "
@@ -102,3 +134,19 @@ def fetch_url(url: str, connection_name: str) -> FetchUrlResult:
                     body=body,
                     truncated=truncated,
                 )
+
+            location = resp.headers.get("Location") or resp.headers.get("location")
+            if not location:
+                raise ValueError(
+                    f"HTTP GET returned {resp.status_code} with no Location header at {url}"
+                )
+            next_url = urljoin(url, location)
+            logger.debug(
+                f"fetch_url redirect url={url} status={resp.status_code} to={next_url}"
+            )
+            _validate_url_host(next_url, conn.url_allowlist)
+            url = next_url
+
+    raise ValueError(
+        f"fetch_url exceeded {_MAX_REDIRECTS} redirects (last url: {url})"
+    )

spark_executor/tools/requests.py

@@ -504,6 +504,8 @@ class ListApplicationsRequest(BaseModel):
     )
     limit: int = Field(
         default=100,
+        ge=1,
+        le=10000,
         description=(
             "Maximum number of applications to return. YARN has no "
             "offset-based pagination, so for large clusters use state/queue "

tests/unit/test_connection_tools.py

@@ -121,6 +121,15 @@ def test_delete_connection_returns_status(_fresh_store):
 def test_delete_connection_unknown_raises(_fresh_store):
     with pytest.raises(KeyError):
         connections.delete_connection("missing")
+
+
+def test_save_connection_preserves_url_allowlist_on_existing_record(_fresh_store):
+    connections.save_connection(name="prod", master="yarn", url_allowlist=["ccam*"])
+    out = connections.save_connection(name="prod", master="spark://new:7077")
+    assert out["master"] == "spark://new:7077"
+    assert out["url_allowlist"] == ["ccam*"]
+
+
 def test_update_connection_changes_specified_field(_fresh_store):
     connections.save_connection(name="prod", master="yarn")
     out = connections.update_connection(name="prod", master="spark://new:7077")

tests/unit/test_external_jobs.py

@@ -5,6 +5,7 @@ from unittest.mock import patch
 import pytest
 
 from spark_executor.core import connection_store
+from spark_executor.core.yarn_client import YarnError
 from spark_executor.models import ApplicationSummary, Connection
 from spark_executor.tools import connections, external_jobs
 from spark_executor.tools.requests import ListApplicationsRequest
@@ -290,3 +291,15 @@ def test_list_applications_maps_yarn_json_to_summary(fresh_stores):
     assert summary.finished_time == 0
     assert summary.tracking_url == "http://rm:8088/proxy/application_42"
     assert summary.progress == 12.5
+
+
+def test_list_applications_raises_on_404(fresh_stores):
+    external_jobs.conn_store.save(
+        Connection(name="prod", master="yarn", yarn_rm_url="http://rm:8088")
+    )
+    with patch(
+        "spark_executor.tools.external_jobs.list_applications_yarn",
+        side_effect=YarnError("YARN list applications failed: 404"),
+    ):
+        with pytest.raises(YarnError, match="YARN list applications failed"):
+            external_jobs.list_applications("prod")

tests/unit/test_fetch_url.py

@@ -4,11 +4,25 @@ from unittest.mock import patch
 
 import httpx
 import pytest
+from pydantic import ValidationError
 
 from spark_executor.core import connection_store
 from spark_executor.core.connection_store import ConnectionStore
 from spark_executor.models import Connection
 from spark_executor.tools import connections, fetch_url
+from spark_executor.tools.requests import ListApplicationsRequest
+
+
+def _stream_cm(resp):
+    """Wrap a response (or fake response) in a context manager for httpx.stream."""
+    class _CM:
+        def __enter__(self):
+            return resp
+
+        def __exit__(self, exc_type, exc, tb):
+            return False
+
+    return _CM()
 
 
 def _fresh_stores(tmp_path, monkeypatch):
@@ -37,7 +51,9 @@ def test_fetch_url_returns_body_and_status(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello", headers={"content-type": "text/html"})
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url("http://nm01.prod.internal:8042/node", "prod")
     assert out.url == "http://nm01.prod.internal:8042/node"
     assert out.status_code == 200
@@ -63,7 +79,9 @@ def test_fetch_url_truncates_body_over_1mb(fresh_stores):
     )
     big = "x" * (fetch_url._MAX_BODY_BYTES + 1)
     resp = httpx.Response(200, text=big)
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp):
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ):
         out = fetch_url.fetch_url("http://nm01.prod.internal/big", "prod")
     assert out.truncated is True
     assert len(out.body) == fetch_url._MAX_BODY_BYTES
@@ -82,7 +100,9 @@ def test_fetch_url_passes_auth_from_connection(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="ok")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         fetch_url.fetch_url("http://nm01.prod.internal:8042/node", "auth")
     auth = m.call_args.kwargs["auth"]
     assert isinstance(auth, httpx.BasicAuth)
@@ -103,26 +123,13 @@ def test_fetch_url_passes_ssl_verify_from_connection(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="ok")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         fetch_url.fetch_url("http://nm01.prod.internal:8042/node", "insecure")
     assert m.call_args.kwargs["verify"] is False
 
 
-def test_fetch_url_follows_redirects(fresh_stores):
-    fetch_url.conn_store.save(
-        Connection(
-            name="prod",
-            master="yarn",
-            yarn_rm_url="http://rm.prod.internal:8088",
-            url_allowlist=["*.prod.internal"],
-        )
-    )
-    resp = httpx.Response(200, text="ok")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
-        fetch_url.fetch_url("http://nm01.prod.internal:8042/node", "prod")
-    assert m.call_args.kwargs["follow_redirects"] is True
-
-
 def test_fetch_url_allows_host_matching_glob_pattern(fresh_stores):
     fetch_url.conn_store.save(
         Connection(
@@ -133,7 +140,9 @@ def test_fetch_url_allows_host_matching_glob_pattern(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url("http://ccam50:8088/foo", "ccam")
     assert out.status_code == 200
     assert out.body == "hello"
@@ -150,7 +159,9 @@ def test_fetch_url_allows_host_matching_any_of_multiple_globs(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url(
             "http://history.prod.internal:18080/api/v1/info", "prod"
         )
@@ -195,7 +206,9 @@ def test_fetch_url_glob_match_does_not_require_suffix_overlap(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url("http://ccam50:8088/", "prod")
     assert out.status_code == 200
     assert out.body == "hello"
@@ -211,7 +224,9 @@ def test_fetch_url_accepts_ip_literal_when_in_url_allowlist(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url("http://10.0.0.1/secret", "prod")
     assert out.status_code == 200
     assert out.body == "hello"
@@ -227,7 +242,9 @@ def test_fetch_url_accepts_https_when_in_url_allowlist(fresh_stores):
         )
     )
     resp = httpx.Response(200, text="hello")
-    with patch("spark_executor.tools.fetch_url.httpx.get", return_value=resp) as m:
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream", return_value=_stream_cm(resp)
+    ) as m:
         out = fetch_url.fetch_url("https://ccam1.example.com/secure", "prod")
     assert out.status_code == 200
     assert out.body == "hello"
@@ -271,3 +288,117 @@ def test_fetch_url_omitted_url_allowlist_defaults_to_empty_and_rejects(fresh_sto
     )
     with pytest.raises(ValueError, match="is not in Connection.url_allowlist"):
         fetch_url.fetch_url("http://nm.prod.internal/", "prod")
+
+
+def test_fetch_url_rejects_redirect_to_disallowed_host(fresh_stores):
+    fetch_url.conn_store.save(
+        Connection(
+            name="ccam",
+            master="yarn",
+            yarn_rm_url="http://ccam1:8088",
+            url_allowlist=["ccam*"],
+        )
+    )
+    redirect = httpx.Response(302, headers={"Location": "http://evil.com/"})
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream",
+        return_value=_stream_cm(redirect),
+    ) as m:
+        with pytest.raises(ValueError, match="is not in Connection.url_allowlist"):
+            fetch_url.fetch_url("http://ccam50/foo", "ccam")
+    assert m.call_count == 1
+
+
+def test_fetch_url_follows_redirect_to_allowed_host(fresh_stores):
+    fetch_url.conn_store.save(
+        Connection(
+            name="prod",
+            master="yarn",
+            yarn_rm_url="http://rm.prod.internal:8088",
+            url_allowlist=["*.internal"],
+        )
+    )
+    redirect = httpx.Response(
+        302, headers={"Location": "http://other.internal/"}
+    )
+    final = httpx.Response(200, text="ok")
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream",
+        side_effect=[_stream_cm(redirect), _stream_cm(final)],
+    ) as m:
+        out = fetch_url.fetch_url("http://foo.internal/", "prod")
+    assert out.status_code == 200
+    assert out.body == "ok"
+    assert m.call_count == 2
+    assert m.call_args_list[1].args[1] == "http://other.internal/"
+
+
+def test_fetch_url_rejects_redirect_to_ip_literal_not_in_allowlist(fresh_stores):
+    fetch_url.conn_store.save(
+        Connection(
+            name="ccam",
+            master="yarn",
+            yarn_rm_url="http://ccam1:8088",
+            url_allowlist=["ccam*"],
+        )
+    )
+    redirect = httpx.Response(302, headers={"Location": "http://10.0.0.1/"})
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream",
+        return_value=_stream_cm(redirect),
+    ) as m:
+        with pytest.raises(ValueError, match="is not in Connection.url_allowlist"):
+            fetch_url.fetch_url("http://ccam50/foo", "ccam")
+    assert m.call_count == 1
+
+
+def test_fetch_url_streams_body_and_truncates_without_buffering_full(fresh_stores):
+    """The response body must be read incrementally; .text must not be accessed."""
+
+    class _FakeStreamResponse:
+        def __init__(self, chunks):
+            self.status_code = 200
+            self.headers = httpx.Headers({"content-type": "text/plain"})
+            self.encoding = "utf-8"
+            self._chunks = chunks
+
+        def iter_bytes(self):
+            yield from self._chunks
+
+        @property
+        def text(self):
+            raise AssertionError("response.text should not be accessed")
+
+    fetch_url.conn_store.save(
+        Connection(
+            name="prod",
+            master="yarn",
+            url_allowlist=["*.prod.internal"],
+        )
+    )
+    chunk_size = 400_000
+    chunks = [
+        b"a" * chunk_size,
+        b"b" * chunk_size,
+        b"c" * chunk_size,
+    ]
+    fake = _FakeStreamResponse(chunks)
+    with patch(
+        "spark_executor.tools.fetch_url.httpx.stream",
+        return_value=_stream_cm(fake),
+    ) as m:
+        out = fetch_url.fetch_url("http://nm01.prod.internal/big", "prod")
+    assert m.call_count == 1
+    assert out.truncated is True
+    assert len(out.body) == fetch_url._MAX_BODY_BYTES
+    assert out.body.startswith("a" * chunk_size)
+    # Only the first 200 KB of the third chunk were consumed before truncation.
+    assert out.body[-200_000:] == "c" * 200_000
+
+
+def test_list_applications_request_limit_bounds():
+    assert ListApplicationsRequest(connection_name="prod", limit=10000).limit == 10000
+    with pytest.raises(ValidationError):
+        ListApplicationsRequest(connection_name="prod", limit=0)
+    with pytest.raises(ValidationError):
+        ListApplicationsRequest(connection_name="prod", limit=10001)