Loading ToolBench
MCP server for synthesizing workspace refactor patches using multi-file context graph analysis and AST-aware patch composition
This MCP server has critical definition quality issues across all dimensions. Only one tool is present (execute_skill_action), and while it has a name and basic description, the implementation is severely incomplete. The source code shows only a stub print statement that outputs a hardcoded JSON response, there is no actual tool registration, no schema validation, no error handling, and no real implementation. The tool description is present but minimal (47 chars). Input schema is documented in the evaluation context but not visible in the actual mcp_server.py code shown. The client.py and example_usage.py files appear to be mock implementations, not the actual MCP server logic. The server does not implement the MCP protocol properly, it simply prints a JSON object and exits, making it non-functional as an MCP server.
Execute patch composition and workspace refactoring skill action
Tool schema not visible in source code. The mcp_server.py file does not contain any actual tool schema definitions, only a print statement.
No actual MCP protocol implementation. The mcp_server.py file contains no request/response handling, no tool invocation logic, and no proper MCP server lifecycle. It only prints a static JSON blob and exits.
Tool description is minimal (47 characters). While present, it lacks actionable context for LLM selection. No guidance on when to use, prerequisites, or what happens as a side effect (the tool WRITES files).
No error handling or recovery guidance. Tool modifies workspace (WRITE risk) with no documented error responses, validation rules, or next-step guidance for failures.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 23 | 2026-07-28+ | v2 |
No input validation visible in code. Parameters (instruction, workspace_root) have descriptions in schema but no constraints, format specs, or length limits documented. Code shows no validation logic.
Missing output schema documentation. The tool is described as executing patch composition and refactoring, but no output schema is documented in the MCP server code. Client.py shows mock return values but these are not part of the server contract.
Tool name lacks clarity on scope. 'execute_skill_action' is generic and could apply to many operations. More specific name like 'compose_workspace_patch' or 'apply_refactoring_patch' would better convey intent.
No confirmation or dry-run for destructive operation. Tool modifies workspace files (WRITE risk) with no option to preview changes, confirm before applying, or rollback on failure.