Loading ToolBench
Dual-stack compatibility middleware for MCP servers: answer both the legacy 2025-11-25 handshake and the new 2026-07-28 stateless spec at once, plus a probe that proves both work.
Server provides 7 tools with mixed quality. Tool naming follows verb_noun convention (echo, get-sum, read_text_file, list_directory, create_entities, read_graph, sequentialthinking). Most tools have descriptions (6/7), but several are generic or incomplete. Schemas are present but parameter descriptions vary widely. The 'sequentialthinking' tool has a complex schema with many optional fields that lack clear guidance on when to use each. File system tools (read_text_file, list_directory) have good descriptions explaining constraints. The 'create_entities' tool lacks output schema documentation. Error handling is minimal, no recovery guidance or actionable error messages visible in tool definitions.
Create multiple new entities in the knowledge graph
Echoes back the input string
Returns the sum of two numbers
Get a detailed listing of all files and directories in a specified path. Results distinguish files and directories with [FILE] and [DIR] prefixes. Only works within allowed directories.
Read the entire knowledge graph
Read the complete contents of a file from the file system as text. Use the 'head' parameter to read only the first N lines, or 'tail' for the last N lines. Only works within allowed directories.
A detailed tool for dynamic and reflective problem-solving through thoughts. Each thought can build on, question, or revise previous insights as understanding deepens.
read_graph tool lacks description explaining what 'knowledge graph' contains, its structure, or when to call it. Description is too generic.
create_entities tool has no documented output schema. LLMs cannot plan downstream calls or extract entity IDs for chaining.
sequentialthinking tool has 10 optional parameters (isRevision, revisesThought, branchFromThought, branchId, needsMoreThoughts) with minimal guidance on when each applies. Parameter relationships undocumented.
No error handling guidance visible in tool definitions. Tools do not document what errors are retryable, user-fixable, or fatal, or provide recovery suggestions.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 62 | 2026-07-28+ | v2 |
read_text_file and list_directory mention 'allowed directories' constraint but do not document what happens if a path is outside allowed scope or how to discover allowed paths.