Loading ToolBench
Strict AI code reviewer MCP server powered by Groq — finds bugs, vulnerabilities and security issues
The server implements 7 tools with mostly complete schemas and descriptions. Tool naming is verb-forward and clear (analyze_code, compare_code, explain_code, generate_tests, analyze_file, cache_info, generate_report). Descriptions are present and adequate (ranging 40-200 chars). However, several tools lack complete parameter type documentation in visible code, and output schemas are not formally documented. Error handling is minimal, most tools return JSON error responses but lack recovery guidance. No tool annotations (readOnlyHint/destructiveHint/idempotentHint) are present. The server demonstrates competent definition practices but falls short of production-grade rigor.
Strict analysis of a code fragment using Groq LLM.
Analyzes a whole code file from disk. Automatically detects language by file extension. Large files are split into chunks and analyzed in parallel.
Shows cache statistics or clears the cache.
Compares two versions of code and evaluates whether the change is an improvement. Performs a structured diff analysis: identifies what improved, what regressed, and what changed neutrally. Returns a merge recommendation based on the findings.
Explains what code does - step by step and clearly.
Generates a beautiful HTML report from analyze_code or analyze_file results.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) present. LLMs cannot determine which tools are safe to retry or whether they modify state. generate_report is a WRITE operation but lacks destructiveHint; cache_info is REVERSIBLE but lacks annotation.
Output schemas are not formally documented. Tools return JSON but there is no visible schema definition describing the structure of returned fields (issues, warnings, suggestions, score, stats, etc.). LLMs cannot plan downstream operations without knowing response structure.
Error handling lacks recovery guidance. analyze_file returns error_response() with brief messages like 'File not found' or 'File too large' but does not suggest next steps (e.g. 'try analyze_code with a smaller fragment' for large files). compare_code has no visible error handling.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | 2026-07-28+ | v2 |
Generates tests for the provided code.
cache_info tool name is ambiguous. Does 'cache_info' retrieve information, or does it modify state (clear)? The description clarifies, but naming should be verb_noun. Consider 'get_cache_info' (for stats) or split into 'get_cache_stats' and 'clear_cache'.
No pagination or result limits documented. analyze_file chunks files and analyzes them in parallel, but no maximum result set size is declared. If analysis finds hundreds of issues, the JSON response could be very large. Baseline pattern requires explicit limits and pagination.
generate_report description does not specify supported output formats or what HTML structure is produced. Is the HTML a standalone report, embedded snippet, or dashboard? What CSS/JS is included?