Loading ToolBench
Polyglot dependency analysis for codebases. Detects circular dependencies, extracts metrics, and visualizes architecture for Python, Rust, TypeScript, JavaScript, Java, and Go projects.
Tools have basic descriptions and parameter names, but critical gaps undermine quality. Parameter descriptions are largely missing (only tool names and repo_url/github_token documented; most others lack descriptions). Input schemas visible in mcp.json but lack type definitions and constraints. No output schemas documented. Error handling guidance absent. Tool names follow verb_noun convention (check_, estimate_, analyze_, get_, retrieve_, compare_, download_) but descriptions are generic and lack WHEN/WHY context. No parameter validation rules, enums, or format constraints visible. Composition is reasonable (single responsibility per tool), but parameter documentation is severely incomplete.
Analyze a Git repository for dependency structure and architecture
Get account email, plan type, and current token balance
Compare architecture of two repositories side-by-side
Get presigned S3 URL for downloading complete dependency graph JSON
Estimate token cost for analyzing a repository before submission
Check the status of a running analysis job
Parameter descriptions almost entirely missing. Only repo_url and github_token have descriptions; parameters like 'wait_for_completion', 'include_full_graph', 'pricing_choice', 'questions', 'limit', 'skip', 'job_id', 'artifact' lack any documentation. LLMs cannot infer parameter meaning from names alone.
No input schemas have type definitions. mcp.json lists parameter names but provides no 'type' field (string, boolean, integer, object, array).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 47 | 2026-07-28+ | v2 |
Detect circular dependency chains from an analysis artifact
Get recent analysis job history with pagination
Extract high-level architectural metrics from an analysis artifact
Download and return complete analysis artifact from a completed job
No output schemas documented. Tools return analysis artifacts, job history, metrics, and presigned URLs, but no structured response format is specified. LLMs cannot plan downstream calls or extract required fields without knowing response structure.
No error handling guidance. Tools that call external APIs (pviz backend) can fail, but no recovery hints provided. No indication of retryable vs fatal errors, no actionable error messages documented.
No parameter constraints or enums. Parameters like 'pricing_choice' and 'wait_for_completion' lack enum values or format constraints. LLMs will hallucinate invalid values without explicit constraints.