cutom MCP stdio : Error: predictionsServices.buildChatflow - Error: Invalid MCP Server Config: Error: Security validation failed: Argument contains potential local file access: "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"

[sysam68]

Describe the bug

Hi,

I have forked plantuml MCP server.
Creating an assistant with chatOllama / get-oss

{
  "type": "stdio",
  "command": "node",
  "args": [
    "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"
  ],
  "env": {
    "MCP_TRANSPORT": "stdio",
    "PLANTUML_SERVER_URL": "http://plantuml-server:8080",
    "LOG_LEVEL": "debug"
  }
}

no list of actions

When asking to generate an UML exemple diagram I have the following error:

Error: predictionsServices.buildChatflow - Error: Invalid MCP Server Config: Error: Security validation failed: Argument contains potential local file access: "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"

I have also this environment configuration:

Security configuration for MCP

    - CUSTOM_MCP_SECURITY_CHECKS=false
    - CUSTOM_MCP_PROTOCOL=stdio
    - HTTP_DENY_LIST=
    # Allows usage of all builtin modules
    - TOOL_FUNCTION_BUILTIN_DEP=*
    - ALLOW_BUILTIN_DEP=true

Inspecting the container, all those variables are clearly set.

I have tested my MCP sever with MCP inspector and seems OK, conform to MCP specifications.

flowise logs:
when refreshing actions dropdown list:

{
    "level": "info",
    "message": "⬆️ POST /api/v1/node-load-method/customMCP",
    "package": "server",
    "request": {
        "body": {
            "baseClasses": [
                "Tool"
            ],
            "category": "Tools (MCP)",
            "description": "Custom MCP Config",
            "documentation": "https://github.com/modelcontextprotocol/servers/tree/main/src/brave-search",
            "filePath": "/usr/local/lib/node_modules/flowise/node_modules/flowise-components/dist/nodes/tools/MCP/CustomMCP/CustomMCP.js",
            "icon": "/usr/local/lib/node_modules/flowise/node_modules/flowise-components/dist/nodes/tools/MCP/CustomMCP/customMCP.png",
            "id": "customMCP_0",
            "inputAnchors": [],
            "inputParams": [
                {
                    "display": true,
                    "hideCodeExecute": true,
                    "hint": {
                        "label": "How to use",
                        "value": "\nYou can use variables in the MCP Server Config with double curly braces `{{ }}` and prefix `$vars.<variableName>`. \n\nFor example, you have a variable called \"var1\":\n```json\n{\n    \"command\": \"docker\",\n    \"args\": [\n        \"run\",\n        \"-i\",\n        \"--rm\",\n        \"-e\", \"API_TOKEN\"\n    ],\n    \"env\": {\n        \"API_TOKEN\": \"{{$vars.var1}}\"\n    }\n}\n```\n\nFor example, when using SSE, you can use the variable \"var1\" in the headers:\n```json\n{\n    \"url\": \"https://api.example.com/endpoint/sse\",\n    \"headers\": {\n        \"Authorization\": \"Bearer {{$vars.var1}}\"\n    }\n}\n```\n"
                    },
                    "id": "customMCP_0-input-mcpServerConfig-code",
                    "label": "MCP Server Config",
                    "name": "mcpServerConfig",
                    "placeholder": "{\n    \"command\": \"npx\",\n    \"args\": [\"-y\", \"@modelcontextprotocol/server-filesystem\", \"/path/to/allowed/files\"]\n}",
                    "type": "code"
                },
                {
                    "display": true,
                    "id": "customMCP_0-input-mcpActions-asyncMultiOptions",
                    "label": "Available Actions",
                    "loadMethod": "listActions",
                    "name": "mcpActions",
                    "refresh": true,
                    "type": "asyncMultiOptions"
                }
            ],
            "inputs": {
                "mcpActions": "[\"decode_plantuml\",\"encode_plantuml\",\"generate_plantuml_diagram\"]",
                "mcpServerConfig": "{\n  \"type\": \"stdio\",\n  \"command\": \"node\",\n  \"args\": [\n    \"/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js\"\n  ],\n  \"env\": {\n    \"MCP_TRANSPORT\": \"stdio\",\n    \"PLANTUML_SERVER_URL\": \"http://plantuml-server:8080\",\n    \"LOG_LEVEL\": \"DEBUG\",\n    \"MCP_API_KEY\": \"1234\"\n  }\n}"
            },
            "label": "Custom MCP",
            "loadMethod": "listActions",
            "loadMethods": {},
            "name": "customMCP",
            "outputAnchors": [
                {
                    "description": "Custom MCP Config",
                    "id": "customMCP_0-output-customMCP-Tool",
                    "label": "Custom MCP Tool",
                    "name": "customMCP",
                    "type": "Tool"
                }
            ],
            "outputs": {},
            "type": "Custom MCP Tool",
            "version": 1.1
        },
        "headers": {
            "accept": "application/json, text/plain, */*",
            "accept-encoding": "gzip, deflate, br",
            "accept-language": "fr-FR,fr;q=0.9",
            "content-length": "2373",
            "content-type": "application/json",
            "cookie": "",
            "host": "ob-flow.fmpn.fr",
            "origin": "",
            "priority": "u=3, i",
            "referer": "https://******/assistants/custom/a51990ec-449d-479b-8348-11fc767d143f",
            "remote-email": "****",
            "remote-groups": "admins",
            "remote-name": "***",
            "remote-user": "****",
            "sec-fetch-dest": "empty",
            "sec-fetch-mode": "cors",
            "sec-fetch-site": "same-origin",
            "user-agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/18.6 Safari/605.1.15",
            "x-forwarded-for": "****",
            "x-forwarded-host": "****",
            "x-forwarded-port": "443",
            "x-forwarded-proto": "https",
            "x-forwarded-server": "2a1aa7ab7502",
            "x-real-ip": "*****",
            "x-request-from": "internal"
        },
        "method": "POST",
        "params": {},
        "query": {},
        "url": "/api/v1/node-load-method/customMCP"
    },
    "timestamp": "2025-10-29 08:38:09"
}
2025-10-29 08: 38: 09 [INFO
]: ⬆️ POST /api/v1/node-load-method/customMCP

when asking to generate a UML exemple diagram:

2025-10-29 08:32:04 [ERROR]: [server]:70ba694b-451a-4124-a9fd-ea0cfcdf24a5/a51990ec-449d-479b-8348-11fc767d143f/5756a90a-2941-4f93-aaf3-5e03c622d4d5 Error: Error: Invalid MCP Server Config: Error: Security validation failed: Argument contains potential local file access: "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"
Error: Error: Invalid MCP Server Config: Error: Security validation failed: Argument contains potential local file access: "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"
    at buildFlow (/usr/local/lib/node_modules/flowise/dist/utils/index.js:550:19)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async executeFlow (/usr/local/lib/node_modules/flowise/dist/utils/buildChatflow.js:384:28)
    at async utilBuildChatflow (/usr/local/lib/node_modules/flowise/dist/utils/buildChatflow.js:871:28)
    at async Object.buildChatflow (/usr/local/lib/node_modules/flowise/dist/services/predictions/index.js:9:28)
    at async createPrediction (/usr/local/lib/node_modules/flowise/dist/controllers/predictions/index.js:72:41)

To Reproduce

1. mount a folder on Flowise official container under /mcp-server
2. in this folder git clone https://github.com/sysam68/plantuml-mcp-server.git
3. cd plantuml-mcp-server
4. test with node /dist/plantuml-mcp-server.js
5. create a custom assistant, add custom MCP tool and copy the MCP configuration:

{
  "type": "stdio",
  "command": "node",
  "args": [
    "/mcp-servers/plantuml-mcp-server/dist/plantuml-mcp-server.js"
  ],
  "env": {
    "MCP_TRANSPORT": "stdio",
    "PLANTUML_SERVER_URL": "http://plantuml-server:8080",
    "LOG_LEVEL": "debug"
  }
}

6. no action list appears
7. check the log in Flowise

Expected behavior

having a list and execution

Screenshots

No response

Flow

No response

Use Method

None

Flowise Version

No response

Operating System

None

Browser

None

Additional context

No response

[sysam68]

Flowise version 3.0.8, tested in mode queue and main.

[HenryHengZJ]

Its related to how you setup MCP.
Most are using npx.
If you are using node , you need to point the executable node file path. for begineer, i recommend just use the stremable http mcp https://docs.flowiseai.com/tutorials/tools-and-mcp#custom-mcp

[sysam68]

Thank you for the Link... It is the post I have followed.

Have a look at #5389
Issue with sse.

[sysam68]

Its related to how you setup MCP.

Most are using npx.

If you are using node , you need to point the executable node file path. for begineer, i recommend just use the stremable http mcp https://docs.flowiseai.com/tutorials/tools-and-mcp#custom-mcp

Hello, @HenryHengZJ any update on the issue ?

[sysam68]

Just tested with context7 both npx and http both fails:
{
"command": "npx",
"args": ["-y", "@upstash/context7-mcp"]
}

get documentation on plantuml
Error: predictionsServices.buildChatflow - json: cannot unmarshal string into Go struct field ChatRequest.tools of type api.Tool

and with http:
get documentation from context7 on plantuml
Error: predictionsServices.buildChatflow - json: cannot unmarshal string into Go struct field ChatRequest.tools of type api.Tool

[sysam68]

Tested with my last version of MCP server, directly from Flowise official docker:
/mcp-servers/plantuml-mcp-server # printf '{"jsonrpc":"2.0","id":1,"method":"tools/list"}\n' | npx -y /mcp-servers/plantuml-mcp-server

[2025-10-31T22:42:53.328Z] [INFO] Log level set to debug
[2025-10-31T22:42:53.329Z] [WARNING] MCP_API_KEY not set. Server will accept unauthenticated requests.
[2025-10-31T22:42:53.330Z] [INFO] PlantUML MCP server running on stdio transport
[2025-10-31T22:42:53.331Z] [DEBUG] ListTools request received
{
    "result": {
        "tools": [
            {
                "name": "generate_plantuml_diagram",
                "title": "Generate PlantUML Diagram",
                "description": "Generate a PlantUML diagram with syntax validation. Returns diagram URLs on success or structured errors for auto-fix workflows.",
                "inputSchema": {
                    "type": "object",
                    "properties": {
                        "plantuml_code": {
                            "type": "string",
                            "description": "PlantUML diagram code that will be validated and rendered."
                        },
                        "format": {
                            "type": "string",
                            "enum": [
                                "svg",
                                "png"
                            ],
                            "default": "svg",
                            "description": "Output image format."
                        }
                    },
                    "required": [
                        "plantuml_code"
                    ]
                },
                "outputSchema": {
                    "type": "object",
                    "properties": {
                        "success": {
                            "type": "boolean"
                        },
                        "format": {
                            "type": "string",
                            "enum": [
                                "svg",
                                "png"
                            ]
                        },
                        "diagram_url": {
                            "type": "string",
                            "format": "uri"
                        },
                        "markdown_embed": {
                            "type": "string"
                        },
                        "encoded_diagram": {
                            "type": "string"
                        },
                        "validation_failed": {
                            "type": "boolean"
                        },
                        "error_details": {
                            "type": "object",
                            "properties": {
                                "error_message": {
                                    "type": "string"
                                },
                                "error_line": {
                                    "type": "integer"
                                },
                                "problematic_code": {
                                    "type": "string"
                                },
                                "full_plantuml": {
                                    "type": "string"
                                },
                                "full_context": {
                                    "type": "string"
                                }
                            }
                        },
                        "retry_instructions": {
                            "type": "string"
                        },
                        "error_message": {
                            "type": "string"
                        }
                    },
                    "required": [
                        "success"
                    ]
                }
            },
            {
                "name": "encode_plantuml",
                "title": "Encode PlantUML",
                "description": "Encode PlantUML code for usage in URLs or PlantUML servers.",
                "inputSchema": {
                    "type": "object",
                    "properties": {
                        "plantuml_code": {
                            "type": "string",
                            "description": "PlantUML diagram code to encode."
                        }
                    },
                    "required": [
                        "plantuml_code"
                    ]
                },
                "outputSchema": {
                    "type": "object",
                    "properties": {
                        "success": {
                            "type": "boolean"
                        },
                        "encoded": {
                            "type": "string"
                        },
                        "svg_url": {
                            "type": "string",
                            "format": "uri"
                        },
                        "png_url": {
                            "type": "string",
                            "format": "uri"
                        },
                        "error_message": {
                            "type": "string"
                        }
                    },
                    "required": [
                        "success"
                    ]
                }
            },
            {
                "name": "decode_plantuml",
                "title": "Decode PlantUML",
                "description": "Decode an encoded PlantUML string back to PlantUML source.",
                "inputSchema": {
                    "type": "object",
                    "properties": {
                        "encoded_string": {
                            "type": "string",
                            "description": "Encoded PlantUML string to decode."
                        }
                    },
                    "required": [
                        "encoded_string"
                    ]
                },
                "outputSchema": {
                    "type": "object",
                    "properties": {
                        "success": {
                            "type": "boolean"
                        },
                        "decoded": {
                            "type": "string"
                        },
                        "error_message": {
                            "type": "string"
                        }
                    },
                    "required": [
                        "success"
                    ]
                }
            }
        ]
    },
    "jsonrpc": "2.0",
    "id": 1
}

And from Flowise, the MCP server is not started either with npc or node:
{ "level": "info", "message": "⬆️ POST /api/v1/node-load-method/customMCP", "package": "server", "request": { "body": { "baseClasses": [ "Tool" ], "category": "Tools (MCP)", "description": "Custom MCP Config", "documentation": "https://github.com/modelcontextprotocol/servers/tree/main/src/brave-search", "filePath": "/usr/local/lib/node_modules/flowise/node_modules/flowise-components/dist/nodes/tools/MCP/CustomMCP/CustomMCP.js", "icon": "/usr/local/lib/node_modules/flowise/node_modules/flowise-components/dist/nodes/tools/MCP/CustomMCP/customMCP.png", "id": "customMCP_0", "inputAnchors": [], "inputParams": [ { "display": true, "hideCodeExecute": true, "hint": { "label": "How to use", "value": "\nYou can use variables in the MCP Server Config with double curly braces {{ }}and prefix$vars.. \n\nFor example, you have a variable called \"var1\":\n```json\n{\n \"command\": \"docker\",\n \"args\": [\n \"run\",\n \"-i\",\n \"--rm\",\n \"-e\", \"API_TOKEN\"\n ],\n \"env\": {\n \"API_TOKEN\": \"{{$vars.var1}}\"\n }\n}\n```\n\nFor example, when using SSE, you can use the variable \"var1\" in the headers:\n```json\n{\n \"url\": \"https://api.example.com/endpoint/sse\",\n \"headers\": {\n \"Authorization\": \"Bearer {{$vars.var1}}\"\n }\n}\n```\n" }, "id": "customMCP_0-input-mcpServerConfig-code", "label": "MCP Server Config", "name": "mcpServerConfig", "placeholder": "{\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@modelcontextprotocol/server-filesystem\", \"/path/to/allowed/files\"]\n}", "type": "code" }, { "display": true, "id": "customMCP_0-input-mcpActions-asyncMultiOptions", "label": "Available Actions", "loadMethod": "listActions", "name": "mcpActions", "refresh": true, "type": "asyncMultiOptions" } ], "inputs": { "mcpActions": "", "mcpServerConfig": "{\n \"command\": \"npx\",\n \"args\": [\"-y\", \n \"/mcp-servers/plantuml-mcp-server\"]\n}" }, "label": "Custom MCP", "loadMethod": "listActions", "loadMethods": {}, "name": "customMCP", "outputAnchors": [ { "description": "Custom MCP Config", "id": "customMCP_0-output-customMCP-Tool", "label": "Custom MCP Tool", "name": "customMCP", "type": "Tool" } ], "outputs": {}, "type": "Custom MCP Tool", "version": 1.1 }, "headers": { "accept": "application/json, text/plain, */*", "accept-encoding": "gzip, deflate, br", "accept-language": "fr-FR,fr;q=0.9", "content-length": "2085", "content-type": "application/json", "cookie": "authelia_session=gbH$lGAcVSb8%0lGJTbA!EDjjZ$RBky3; token=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6Ijk4ODY4Y2ZmLTdlMTgtNDlmZS05N2ZhLTFiYmMyY2MwNmUwOCIsInVzZXJuYW1lIjoiU3lsdmFpbiBNT0lORSIsIm1ldGEiOiI0NTBkN2Q3MGNjOTI2M2IxN2QwZjRiZWM4MTlmOTg3Zjo3MDQ3ZjQwZmZiZWJiOTYwNjU1YTY0ODEzNmZmY2E0MmZkYWNkNTk2MTQ0MTg5NmVhYmUyMGQ4MDI4ZDIxY2UyODVlYzhhODY0MTIzY2M4MjJhMTVkOWQ0ODE2NjgxNDkzYjQ1YWU3ZmFmY2NiNzZhNzE3Njk1MWFiMzg0YWU3YWZkNTI5ZmJkZmViZWQ1YWQ2ZmY0OWI2ZmY0M2ZhOGQ1IiwiaWF0IjoxNzYxOTQxMTM2LCJuYmYiOjE3NjE5NDExMzYsImV4cCI6MTc2MTk2MjczNiwiYXVkIjoiQVVESUVOQ0UiLCJpc3MiOiJJU1NVRVIifQ.MeTi1yq7Mvo07bMrugXZk2xNtIV4lskawmpAc0_1YDU; connect.sid=s%3AzDjlN-WtahSAanOzT_5FN7aUViKQ78jZ.bNbouvy%2BIUd43sXnPn5yi1yx6x4B%2BO3j26Nmt%2Bccku0; refreshToken=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpZCI6Ijk4ODY4Y2ZmLTdlMTgtNDlmZS05N2ZhLTFiYmMyY2MwNmUwOCIsInVzZXJuYW1lIjoiU3lsdmFpbiBNT0lORSIsIm1ldGEiOiIyMzEwMjI2ZjEyOTE1NTU5ODBiNjZlYzgzZDY1ZmQ1YzpiZDdkNzRiNzYzODJkOTAxZTI5ZDhiZjQ4MmZhNGFjNjFkMDI3ZjJiM2UxYzFlZWUzMjNjNGY3OGUzNzNkNTI1NGZkZWUzNjQ3NjQ5NWQxNGMwNDc3YTYzNTE4NWE0NjJiZGY5Y2RiYTRkYzNlMGNkMDllYTZiNWMyMWM4YmU4YTZiMmVmNzY5MGZjOTk2OGRmYjRlODVjNTA2YzA3YTRiIiwiaWF0IjoxNzYxODI2NTQxLCJuYmYiOjE3NjE4MjY1NDEsImV4cCI6MTc2NDQxODU0MSwiYXVkIjoiQVVESUVOQ0UiLCJpc3MiOiJJU1NVRVIifQ.g0fWGVjrehabf1GmvvU6-RGQW1f-c-iRk1SYyT_lC_U", "host": "******", "origin": "https://*****", "priority": "u=3, i", "referer": "https://o*****/assistants/custom/a51990ec-449d-479b-8348-11fc767d143f", "remote-email": "******", "remote-groups": "****", "remote-name": "****", "remote-user": "****", "sec-fetch-dest": "empty", "sec-fetch-mode": "cors", "sec-fetch-site": "same-origin", "user-agent": "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/605.1.15 (KHTML, like Gecko) Version/18.6 Safari/605.1.15", "x-forwarded-for": "1****", "x-forwarded-host": "*****", "x-forwarded-port": "443", "x-forwarded-proto": "https", "x-forwarded-server": "2a1aa7ab7502", "x-real-ip": "172.18.0.1", "x-request-from": "internal" }, "method": "POST", "params": {}, "query": {}, "url": "/api/v1/node-load-method/customMCP" }, "timestamp": "2025-10-31 22:49:14" }

and with sse:
Error: predictionsServices.buildChatflow - template: :108:130: executing "" at <index $prop.Type 0>: error calling index: reflect: slice index out of range

test with MCP inspector show that everything is fine.

Asked GPT to test compatibility with MCP standard : 98%
why is context7 not working to ? either with npx or http ?
Have also tested Flowise 3.0.6, 3.0.7 and 3.08 : same result.
MCP for Postgres works.