Loading ToolBench
A Grey Hack MCP server with GitHub code search, Greybel-JS transpilation, API validation and script generation
Server has 4 tools with basic schemas and descriptions, but significant quality gaps. All tools have descriptions (10-80 chars, mostly short), and input schemas are present with type definitions. However, descriptions are generic and lack actionable context for LLM selection. Parameter descriptions are minimal. Output schemas are not documented. Error handling returns generic error strings without recovery guidance. No tool annotations (readOnlyHint, destructiveHint, idempotentHint). Tool names follow verb_noun pattern but lack clarity on distinctions (e.g., 'validate' vs 'transpile' both operate on code). The transpile_greyscript tool has a placeholder implementation that does not actually transpile, undermining its credibility.
Generate a GreyScript code template for common Grey Hack game tasks
Search for Grey Hack code examples on GitHub
Transpile GreyScript code to JavaScript using Greybel-JS
Validate GreyScript code against the official API documentation
Output schemas not documented. LLMs cannot plan downstream calls or extract required fields. search_greyhack_code returns {results, total_count} but structure of result items is inferred, not declared. validate_greyscript returns {valid, warnings, errors, api_calls} without field type definitions.
Error handling lacks recovery guidance. Errors return generic strings like 'Error searching GitHub: {error.message}' without actionable next steps. Pattern requires: 'User not found. Try search_users() with a partial name.' Current errors give LLM nothing to act on.
Tool descriptions are too short (40-80 chars) and lack context for LLM selection. 'Search for Grey Hack code examples on GitHub' does not explain WHEN to use this vs other discovery tools, what structure is returned, or prerequisites (GITHUB_TOKEN required). Baseline: 194 chars average for A+ tools.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
transpile_greyscript has placeholder implementation that does not actually transpile. Handler returns simulated output with comment '// Transpiled code would go here' and hardcoded template. Tool is non-functional and misleading.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All 4 tools are read-only but this is not declared in schema. Agents cannot determine which tools are safe to retry or have side effects.