Loading ToolBench
FileMaker MCP server for analyzing and managing FileMaker database projects, providing tools for schema exploration, file management, and reference tracking
kibuild-mcp provides 17 tools with basic schemas and descriptions, but falls short of production quality. All tools have descriptions (10-150 chars, mostly adequate) and input schemas with type definitions. However, descriptions lack LLM-optimized guidance on WHEN to use each tool vs. alternatives. Output schemas are not documented, callers cannot predict response structure. Error handling is minimal; most tools return raw Go errors without recovery guidance. Tool naming is verb-forward (list_, read_, write_, find_, get_) but several reference tools (find_layout_references_to_scripts, find_script_references_in_layouts) are confusingly similar and could be consolidated. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) despite clear risk levels. Security-sensitive tools like write_file and generate_schema_map lack confirmation patterns or dry-run support.
Export database schema from FileMaker via plugin connection
Find all calculation fields that reference specified fields
Find all layouts that reference specified scripts via triggers or buttons
Find all layouts that reference specified table occurrences
Find all layouts that reference specified value lists
Find all layouts that invoke specified scripts via triggers or buttons
Find all scripts that call specified scripts via Perform Script steps
Output schemas not documented. Tools return unstructured strings or JSON without declaring expected fields, types, or pagination. LLMs cannot predict response structure or chain tools reliably.
Confusing reference-finding tool names. find_layout_references_to_scripts, find_script_references_in_layouts, find_field_references_in_calculations, find_variable_references_in_scripts, and find_to_references are semantically similar and differ only in subtle direction/target combinations. LLMs will struggle to pick the right one.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 57 | 2026-07-28+ | v2 |
Find a table by name within a specified database
Find all references to specified table occurrences across layouts, scripts, and relationships
Find all script steps that reference specified variables
Generate a comprehensive workspace schema map documenting all database structures, tables, layouts, scripts, and table occurrences
Get the currently active FileMaker context (database, layout, table) via plugin connection
List directory contents with file metadata (name, is_dir, size)
Read and return the contents of a file
Read layout definition from FileMaker via plugin connection
Resolve and validate a file path within project boundaries, preventing directory traversal attacks
Write content to a file, creating parent directories if needed
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). write_file and generate_schema_map are destructive but unmarked. Agents cannot distinguish safe reads from risky writes without parsing descriptions.
Minimal error handling. Tools return raw Go errors (e.g., 'no active project directory set', 'security boundary violation') without recovery guidance. LLMs cannot determine if errors are retryable or require user intervention.
Descriptions lack WHEN-to-use guidance. Most descriptions state WHAT the tool does but not when to call it vs. similar tools or what prerequisites exist. E.g., export_schema vs. read_layout both read FileMaker structures but distinction is unclear.