Loading ToolBench
A custom MCP server to perform code reviews
This server has two tools with reasonable naming (verb_noun pattern) and descriptions that clearly state WHEN to use each tool. However, there are significant gaps: parameter descriptions are present but minimal, input schemas lack crucial constraints (enums, min/max), output schemas are not documented, and error handling is basic. The tool names in the actual registration code differ from the provided specs (analyze_repo → analyze_codebase_structure, code_review → comprehensive_quality_review), indicating possible inconsistency. Tool descriptions are well-written for LLM decision-making (~150-170 chars), meeting the 10-1024 char baseline, but parameter documentation is sparse. No validation of enum constraints at the parameter level, and responses are JSON strings without schema documentation.
Use this tool when you need to analyze a code repository structure without performing a detailed review. This tool flattens the repository into a textual representation and is ideal for getting a high-level overview of code organization, directory structure, and file contents. Use it before code_review when you need to understand the codebase structure first, or when a full code review is not needed.
Use this tool when you need a comprehensive code review with specific feedback on code quality, security issues, performance problems, and maintainability concerns. This tool performs in-depth analysis on a repository or specific files and returns structured results including issues found, their severity, recommendations for fixes, and overall strengths of the codebase. Use it when you need actionable insights to improve code quality or when evaluating a codebase for potential problems.
Parameter descriptions are minimal; repoPath has a description but specificFiles and fileTypes lack meaningful detail about expected values, formats, or constraints.
focusAreas parameter is defined as an enum with ['security', 'performance', 'quality', 'maintainability'], but the description does not explain what each focus area returns or how they differ, leaving LLM uncertain which to choose.
Output schema is not documented. Both tools return JSON-stringified responses, but the structure of the returned object (fields, types, nested objects) is not specified. LLMs cannot reliably extract data without schema documentation.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | C | 63 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 41 | - | v1 |
No pagination or result limit guidance. If analyze_codebase_structure returns a large repository, there is no mechanism to cap results or offer pagination. Tool description states 'flattens the repository into a textual representation' but does not mention limits.
Error handling returns isError: true with plain text descriptions, but does not guide the LLM on recovery. 'Error initializing code review service: ...' is informative but lacks recovery guidance like 'Try checking your LLM_PROVIDER env var or set API keys.'
repoPath parameter is required but its description only says 'Path to the repository to analyze'. Missing clarity on whether this is a local filesystem path, Git URL, or relative/absolute path requirement. LLMs may pass invalid values.