GHSA-4HF8-5MJM-RFGQ

Vulnerability from github – Published: 2026-06-26 21:50 – Updated: 2026-06-26 21:50
VLAI
Summary
Streamable HTTP mode exposes LINE Desktop read/send tools without MCP authentication
Details

Streamable HTTP mode exposes LINE Desktop read/send tools without MCP authentication

Summary

line-desktop-mcp supports a --http-mode Streamable HTTP transport for use with clients such as n8n. In this mode the server binds to 0.0.0.0 and exposes the MCP /mcp endpoint without an MCP-layer authentication check. Any network client that can reach the port can initialize a session, list tools, and call tools that read LINE Desktop chat history or send LINE messages through the already logged-in desktop application.

This is High for deployments where the HTTP port is reachable beyond the local host, because the server acts with the user authority of the logged-in LINE Desktop session. It is lower if the listener is strictly firewalled to trusted local clients.

Affected version

Repository: dtwang/line-desktop-mcp

Current source checked: fbed0d2d3048e63f48a356a1267ed8ec5e78f3ae on main, committed 2026-05-14.

Published npm package checked: line-desktop-mcp@1.1.1.

Source evidence

README.md documents Streamable HTTP mode:

npx line-desktop-mcp@latest --http-mode --port 3000

The same README documents MCP endpoints at /mcp and explains that this mode is intended for clients such as n8n.

src/server.js registers LINE Desktop tools including:

  • get_line_chatroom_history_default
  • get_line_chatroom_history_long
  • get_line_chatroom_history_short
  • send_message_manual
  • send_message_auto

Those tool handlers call into the desktop automation layer: getChatHistory(...) and sendChatMessage(...).

In HTTP mode, src/server.js creates an Express app and Streamable HTTP transport, accepts POSTs to /mcp, creates sessions, connects the transport to the MCP server, and calls transport.handleRequest(...). I did not find an authentication or bearer-token check before session creation or tool invocation.

The listener is explicitly network-bound:

app.listen(port, 0.0.0.0, () => {
  console.error(`LINE Desktop MCP Server running on Streamable HTTP mode`);
  console.error(`  Local:   http://127.0.0.1:${port}${endpoint}`);
  console.error(`  Network: http://0.0.0.0:${port}${endpoint}`);
});

Vulnerability chain

  1. A user starts the server with --http-mode --port 3000.
  2. The server binds on 0.0.0.0:3000, not only loopback.
  3. A network client reaches /mcp and sends the normal MCP initialize request.
  4. The server creates a Streamable HTTP session without authenticating the caller.
  5. The caller can list and invoke LINE Desktop tools.
  6. Tool calls execute through the logged-in LINE Desktop application on the user workstation.

Impact

An unauthenticated network client can read LINE chat history through the MCP history tools and can send LINE messages through the send-message tools, including send_message_auto when the tool call requests immediate sending. The attacker does not need LINE credentials or a LINE API token; they only need network reachability to the MCP HTTP port.

The practical impact is disclosure of private LINE conversations and unauthorized messages sent as the logged-in desktop user.

Suggested fix

Require authentication before accepting Streamable HTTP MCP sessions or tool calls. For example:

  • require a bearer token or local secret when --http-mode is used;
  • bind HTTP mode to 127.0.0.1 by default unless the operator explicitly opts into network exposure;
  • refuse to start 0.0.0.0 HTTP mode without authentication;
  • document that host.docker.internal / n8n setups must still authenticate to the MCP server.

A defense-in-depth improvement would also keep send_message_auto disabled unless explicitly enabled by a server-side flag, because it converts MCP tool access into immediate message sending as the desktop user.

Severity
8.8 (High)
CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:L/VA:N/SC:N/SI:N/SA:N
Show details on source website
To clipboard 

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 1.1.1"
      },
      "package": {
        "ecosystem": "npm",
        "name": "line-desktop-mcp"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "1.1.2"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-49357"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-306",
      "CWE-862"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-06-26T21:50:49Z",
    "nvd_published_at": "2026-06-19T14:16:23Z",
    "severity": "HIGH"
  },
  "details": "# Streamable HTTP mode exposes LINE Desktop read/send tools without MCP authentication\n\n## Summary\n\n`line-desktop-mcp` supports a `--http-mode` Streamable HTTP transport for use with clients such as n8n. In this mode the server binds to `0.0.0.0` and exposes the MCP `/mcp` endpoint without an MCP-layer authentication check. Any network client that can reach the port can initialize a session, list tools, and call tools that read LINE Desktop chat history or send LINE messages through the already logged-in desktop application.\n\nThis is High for deployments where the HTTP port is reachable beyond the local host, because the server acts with the user authority of the logged-in LINE Desktop session. It is lower if the listener is strictly firewalled to trusted local clients.\n\n## Affected version\n\nRepository: `dtwang/line-desktop-mcp`\n\nCurrent source checked: `fbed0d2d3048e63f48a356a1267ed8ec5e78f3ae` on `main`, committed 2026-05-14.\n\nPublished npm package checked: `line-desktop-mcp@1.1.1`.\n\n## Source evidence\n\n`README.md` documents Streamable HTTP mode:\n\n```text\nnpx line-desktop-mcp@latest --http-mode --port 3000\n```\n\nThe same README documents MCP endpoints at `/mcp` and explains that this mode is intended for clients such as n8n.\n\n`src/server.js` registers LINE Desktop tools including:\n\n- `get_line_chatroom_history_default`\n- `get_line_chatroom_history_long`\n- `get_line_chatroom_history_short`\n- `send_message_manual`\n- `send_message_auto`\n\nThose tool handlers call into the desktop automation layer: `getChatHistory(...)` and `sendChatMessage(...)`.\n\nIn HTTP mode, `src/server.js` creates an Express app and Streamable HTTP transport, accepts POSTs to `/mcp`, creates sessions, connects the transport to the MCP server, and calls `transport.handleRequest(...)`. I did not find an authentication or bearer-token check before session creation or tool invocation.\n\nThe listener is explicitly network-bound:\n\n```js\napp.listen(port, 0.0.0.0, () =\u003e {\n  console.error(`LINE Desktop MCP Server running on Streamable HTTP mode`);\n  console.error(`  Local:   http://127.0.0.1:${port}${endpoint}`);\n  console.error(`  Network: http://0.0.0.0:${port}${endpoint}`);\n});\n```\n\n## Vulnerability chain\n\n1. A user starts the server with `--http-mode --port 3000`.\n2. The server binds on `0.0.0.0:3000`, not only loopback.\n3. A network client reaches `/mcp` and sends the normal MCP initialize request.\n4. The server creates a Streamable HTTP session without authenticating the caller.\n5. The caller can list and invoke LINE Desktop tools.\n6. Tool calls execute through the logged-in LINE Desktop application on the user workstation.\n\n## Impact\n\nAn unauthenticated network client can read LINE chat history through the MCP history tools and can send LINE messages through the send-message tools, including `send_message_auto` when the tool call requests immediate sending. The attacker does not need LINE credentials or a LINE API token; they only need network reachability to the MCP HTTP port.\n\nThe practical impact is disclosure of private LINE conversations and unauthorized messages sent as the logged-in desktop user.\n\n## Suggested fix\n\nRequire authentication before accepting Streamable HTTP MCP sessions or tool calls. For example:\n\n- require a bearer token or local secret when `--http-mode` is used;\n- bind HTTP mode to `127.0.0.1` by default unless the operator explicitly opts into network exposure;\n- refuse to start `0.0.0.0` HTTP mode without authentication;\n- document that `host.docker.internal` / n8n setups must still authenticate to the MCP server.\n\nA defense-in-depth improvement would also keep `send_message_auto` disabled unless explicitly enabled by a server-side flag, because it converts MCP tool access into immediate message sending as the desktop user.",
  "id": "GHSA-4hf8-5mjm-rfgq",
  "modified": "2026-06-26T21:50:50Z",
  "published": "2026-06-26T21:50:49Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/dtwang/line-desktop-mcp/security/advisories/GHSA-4hf8-5mjm-rfgq"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-49357"
    },
    {
      "type": "WEB",
      "url": "https://github.com/dtwang/line-desktop-mcp/commit/680617894981ea93f8f6ceb51ecde7519754d501"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/dtwang/line-desktop-mcp"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:L/VA:N/SC:N/SI:N/SA:N",
      "type": "CVSS_V4"
    }
  ],
  "summary": "Streamable HTTP mode exposes LINE Desktop read/send tools without MCP authentication"
}



Log in or create an account to share your comment.




Tags
Taxonomy of the tags.





Forecast uses a logistic model when the trend is rising, or an exponential decay model when the trend is falling. Fitted via linearized least squares.

Sightings

Author Source Type Date Other

Nomenclature

  • Seen: The vulnerability was mentioned, discussed, or observed by the user.
  • Confirmed: The vulnerability has been validated from an analyst's perspective.
  • Published Proof of Concept: A public proof of concept is available for this vulnerability.
  • Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
  • Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
  • Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
  • Not confirmed: The user expressed doubt about the validity of the vulnerability.
  • Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.



Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.