Failed to validate tool 9f1_read_neo4j_cypher: TypeError: Cannot use 'in' operator to search for 'type' in true

[kunalgokheTSPL]

Expected Behavior (Mandatory)

Failed to validate tool read_neo4j_cypher: TypeError: Cannot use 'in' operator to search for 'type' in true

Actual Behavior (Mandatory)

Failed to validate tool 9f1_read_neo4j_cypher: TypeError: Cannot use 'in' operator to search for 'type' in true

I've used MCP with vscode copilot, It does works with get_schema_cypher functionality, but when i tried by calling read_neo4j_cypher it gives the error "Failed to validate tool 9f1_read_neo4j_cypher"

[a-s-g93]

please provide more information according to the issue template

[kunalgokheTSPL]

please provide more information according to the issue template

Actual Behavior (Mandatory)

Failed to validate tool 9f1_read_neo4j_cypher: TypeError: Cannot use 'in' operator to search for 'type' in true

I've used MCP with vscode copilot, It does works with get_schema_cypher functionality, but when i tried by calling read_neo4j_cypher it gives the error "Failed to validate tool 9f1_read_neo4j_cypher"

[miguelfg]

Very similar to this, I get GitHub Copilot Failed to validate tool 590_read_neo4j_cypher: TypeError: Cannot use 'in' operator to search for 'type' in true . But I wasn't able to return the schema, nor executing any read cypher. I've tried Claude 3.7 Sonnet and GPT 4o

Examples of my prompts:

• use neo4j mcp get_schema_cypher and return the schema of the db
• use neo4j mcp and return the schema of the db
• use neo4j mcp to count the number of Entity nodes
• ...

MCP config as:

            "neo4J": {
                "command": "uvx",
                "args": ["mcp-neo4j-cypher", 
                         "--db-url", "bolt://localhost:7687", 
                         "--password","XXXX"]
            },

It fails so quickly that I doubt the llm is even requested.

[jalakoo]

I ran into the same issue with VS Code. But everything worked fine with VS Code Insiders.

[kaidatar]

Can confirm, this happens to me too.

[jexp]

It would be good if you could also share the generated cypher queries, as in read_cypher it's the cypher query that the LLM generates, so not much we can do from an MCP perspective.

the get_schema tool uses a pre-canned query to determine the schema, which works

[logless]

I have exactly the same issue. Will try VS Code Insiders.

EDIT: confirmed - error disappears when using VSCode Insiders and the MCP server is working properly.