Skip to main content
Glama

json_filter

Filter JSON data from files or URLs using a shape object to extract specific fields and reduce context size for LLM processing.

Instructions

Filter JSON data using a shape object to extract only the fields you want. Provide filePath (local file or HTTP/HTTPS URL) and shape parameters.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
filePathYesPath to the JSON file (local) or HTTP/HTTPS URL to filter
shapeNoShape object (formatted as valid JSON) defining what fields to extract. Use 'true' to include a field, or nested objects for deep extraction. Examples: 1. Extract single field: {"type": true} 2. Extract multiple fields: {"type": true, "version": true, "source": true} 3. Extract nested fields: {"appState": {"gridSize": true, "viewBackgroundColor": true}} 4. Extract from arrays: {"elements": {"type": true, "x": true, "y": true}} - applies to each array item 5. Complex nested extraction: { "type": true, "version": true, "appState": { "gridSize": true, "viewBackgroundColor": true }, "elements": { "type": true, "text": true, "x": true, "y": true, "boundElements": { "type": true, "id": true } } } Note: - Arrays are automatically handled - the shape is applied to each item in the array. - Use json_schema tool to analyse the JSON file schema before using this tool. - Use json_dry_run tool to get a size breakdown of your desired json shape before using this tool.
chunkIndexNoIndex of chunk to retrieve (0-based). If filtered data exceeds 400KB, it will be automatically chunked. Defaults to 0 if not specified.

Implementation Reference

  • Core handler function that ingests JSON from file or URL, parses it, applies shape-based extraction, performs size check and automatic line-based chunking if exceeds 400KB threshold.
    async function processJsonFilter(input: JsonFilterInput): Promise<JsonFilterResult> {
      try {
        // Use strategy pattern to ingest JSON content
        const ingestionResult = await jsonIngestionContext.ingest(input.filePath);
        
        if (!ingestionResult.success) {
          // Map strategy errors to existing error format for backward compatibility
          return {
            success: false,
            error: ingestionResult.error
          };
        }
    
        // Parse JSON
        let parsedData: any;
        try {
          parsedData = JSON.parse(ingestionResult.content);
        } catch (error) {
          return {
            success: false,
            error: {
              type: 'invalid_json',
              message: 'Invalid JSON format in content',
              details: error
            }
          };
        }
    
        // Apply shape filter
        try {
          const filteredData = extractWithShape(parsedData, input.shape);
          
          // Convert filtered data to JSON string to check size
          const filteredJson = JSON.stringify(filteredData, null, 2);
          const filteredSize = new TextEncoder().encode(filteredJson).length;
          
          // Define chunking threshold (400KB)
          const CHUNK_THRESHOLD = 400 * 1024;
          
          // If under threshold, return all data
          if (filteredSize <= CHUNK_THRESHOLD) {
            return {
              success: true,
              filteredData
            };
          }
          
          // Calculate total chunks needed
          const totalChunks = Math.ceil(filteredSize / CHUNK_THRESHOLD);
          const chunkIndex = input.chunkIndex ?? 0; // Default to 0
          
          // Validate chunk index
          if (chunkIndex >= totalChunks || chunkIndex < 0) {
            return {
              success: false,
              error: {
                type: 'validation_error',
                message: `Invalid chunkIndex ${chunkIndex}. Must be between 0 and ${totalChunks - 1}`,
                details: { chunkIndex, totalChunks }
              }
            };
          }
          
          // Line-based chunking
          const lines = filteredJson.split('\n');
          const linesPerChunk = Math.ceil(lines.length / totalChunks);
          const startLine = chunkIndex * linesPerChunk;
          const endLine = Math.min(startLine + linesPerChunk, lines.length);
          
          // Extract chunk as text
          const chunkText = lines.slice(startLine, endLine).join('\n');
          
          return {
            success: true,
            filteredData: chunkText, // Return as string when chunking
            chunkInfo: {
              chunkIndex,
              totalChunks
            }
          };
        } catch (error) {
          return {
            success: false,
            error: {
              type: 'validation_error',
              message: 'Failed to apply shape filter',
              details: error
            }
          };
        }
      } catch (error) {
        return {
          success: false,
          error: {
            type: 'validation_error',
            message: 'Unexpected error during processing',
            details: error
          }
        };
      }
    }
  • src/index.ts:563-681 (registration)
    MCP server registration of the 'json_filter' tool, including detailed description, Zod input schema, and wrapper handler that handles shape parsing, validation, execution via processJsonFilter, and response formatting.
    server.tool(
        "json_filter",
        "Filter JSON data using a shape object to extract only the fields you want. Provide filePath (local file or HTTP/HTTPS URL) and shape parameters.",
        {
            filePath: z.string().describe("Path to the JSON file (local) or HTTP/HTTPS URL to filter"),
            shape: z.unknown().describe(`Shape object (formatted as valid JSON) defining what fields to extract. Use 'true' to include a field, or nested objects for deep extraction.
    
    Examples:
    1. Extract single field: {"type": true}
    2. Extract multiple fields: {"type": true, "version": true, "source": true}
    3. Extract nested fields: {"appState": {"gridSize": true, "viewBackgroundColor": true}}
    4. Extract from arrays: {"elements": {"type": true, "x": true, "y": true}} - applies to each array item
    5. Complex nested extraction: {
       "type": true,
       "version": true,
       "appState": {
         "gridSize": true,
         "viewBackgroundColor": true
       },
       "elements": {
         "type": true,
         "text": true,
         "x": true,
         "y": true,
         "boundElements": {
           "type": true,
           "id": true
         }
       }
    }
    
    Note: 
    - Arrays are automatically handled - the shape is applied to each item in the array.
    - Use json_schema tool to analyse the JSON file schema before using this tool.
    - Use json_dry_run tool to get a size breakdown of your desired json shape before using this tool.
    `),
            chunkIndex: z.number().optional().describe("Index of chunk to retrieve (0-based). If filtered data exceeds 400KB, it will be automatically chunked. Defaults to 0 if not specified.")
        },
        async ({ filePath, shape, chunkIndex }) => {
            try {
                // If shape is a string, parse it as JSON
                let parsedShape = shape;
                if (typeof shape === 'string') {
                    try {
                        parsedShape = JSON.parse(shape);
                    } catch (e) {
                        return {
                            content: [
                                {
                                    type: "text",
                                    text: `Error: Invalid JSON in shape parameter: ${e instanceof Error ? e.message : String(e)}`
                                }
                            ],
                            isError: true
                        };
                    }
                }
    
                
    
                const validatedInput = JsonFilterInputSchema.parse({
                    filePath,
                    shape: parsedShape,
                    chunkIndex
                });
                
                const result = await processJsonFilter(validatedInput);
                
                if (result.success) {
                    // Check if chunking is active
                    if (result.chunkInfo) {
                        // Return chunk data + metadata as separate content items
                        return {
                            content: [
                                {
                                    type: "text",
                                    text: result.filteredData // This is already a string when chunking
                                },
                                {
                                    type: "text",
                                    text: JSON.stringify(result.chunkInfo)
                                }
                            ]
                        };
                    } else {
                        // No chunking - return as normal JSON
                        return {
                            content: [
                                {
                                    type: "text",
                                    text: JSON.stringify(result.filteredData, null, 2)
                                }
                            ]
                        };
                    }
                } else {
                    return {
                        content: [
                            {
                                type: "text",
                                text: `Error: ${result.error.message}`
                            }
                        ],
                        isError: true
                    };
                }
            } catch (error) {
                return {
                    content: [
                        {
                            type: "text",
                            text: `Validation error: ${error instanceof Error ? error.message : String(error)}`
                        }
                    ],
                    isError: true
                };
            }
        }
    );
  • Zod input validation schema used internally by the json_filter handler for filePath, shape, and optional chunkIndex.
    const JsonFilterInputSchema = z.object({
      filePath: z.string().min(1, "File path or HTTP/HTTPS URL is required").refine(
        (val) => val.length > 0 && (val.startsWith('./') || val.startsWith('/') || val.startsWith('http://') || val.startsWith('https://') || !val.includes('/')),
        "Must be a valid file path or HTTP/HTTPS URL"
      ),
      shape: z.any().describe("Shape object defining what to extract"),
      chunkIndex: z.number().int().min(0).optional().describe("Index of chunk to retrieve (0-based)")
    });
  • Recursive helper function that applies the shape filter to JSON data or arrays, extracting only specified fields and nested structures.
    function extractWithShape(data: any, shape: Shape): any {
      if (Array.isArray(data)) {
        return data.map(item => extractWithShape(item, shape));
      }
    
      const result: any = {};
      for (const key in shape) {
        const rule = shape[key];
        if (rule === true) {
          result[key] = data[key];
        } else if (typeof rule === 'object' && data[key] !== undefined) {
          result[key] = extractWithShape(data[key], rule);
        }
      }
      return result;
    }
  • TypeScript type definition for the output of processJsonFilter, including success case with filteredData and optional chunkInfo, or error.
    type JsonFilterResult = {
      readonly success: true;
      readonly filteredData: any;
      readonly chunkInfo?: {
        readonly chunkIndex: number;
        readonly totalChunks: number;
      };
    } | {
      readonly success: false;
      readonly error: JsonSchemaError;
    };
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations provided, the description carries full burden and does well by disclosing key behaviors: automatic chunking when data exceeds 400KB (via chunkIndex parameter explanation), handling of arrays ('Arrays are automatically handled'), and the need for valid JSON shape. It doesn't mention error conditions or performance characteristics, but covers essential operational behavior.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is appropriately sized and front-loaded with the core purpose. The shape parameter examples are extensive but necessary for understanding the tool's capabilities. The sibling tool references are efficiently integrated. Some redundancy exists between the description and schema (filePath explanation).

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 3-parameter tool with no annotations and no output schema, the description provides good coverage: clear purpose, usage guidelines, parameter semantics with examples, and behavioral transparency about chunking and array handling. It doesn't describe the output format or error responses, but given the schema coverage and behavioral details provided, it's mostly complete.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the baseline is 3. The description adds significant value by explaining the shape parameter's purpose ('defining what fields to extract'), providing multiple concrete examples with different use cases, and clarifying array handling behavior. However, it doesn't explain the filePath parameter beyond what's in the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the specific action ('filter JSON data'), the resource ('JSON data'), and the mechanism ('using a shape object to extract only the fields you want'). It distinguishes from sibling tools by mentioning their complementary roles (json_schema for analysis, json_dry_run for size breakdown) rather than overlapping functionality.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description explicitly states when to use this tool ('to extract only the fields you want') and provides clear alternatives: 'Use json_schema tool to analyse the JSON file schema before using this tool' and 'Use json_dry_run tool to get a size breakdown of your desired json shape before using this tool.' This gives the agent specific guidance on tool selection workflow.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Install Server

Other Tools

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/kehvinbehvin/json-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server