Interactive CLI tool for testing remote MCP servers - inspired by Colossus: The Forbin Project, where systems learn to communicate
Forbin is a meta-tool for testing remote MCP servers, not a domain-specific tool suite. Its 3 tools (list_tools, call_tool, get_tool_parameters) are intentionally generic wrappers. However, they suffer from critical definition gaps: descriptions are present but minimal (10-50 chars, below the 50-200 char LLM-optimized baseline); input schemas lack proper type definitions and parameter descriptions; output schemas are undocumented. The call_tool parameters object is untyped and undescribed. Error handling is absent, no guidance on what happens when a tool call fails or parameters are invalid. These gaps are severe for a tool that agents will invoke repeatedly.
Call a tool with the given parameters.
Interactively collect parameters for a tool.
List all available tools from the MCP server.
Input schemas lack type definitions and parameter descriptions. call_tool's 'parameters' field is typed as 'object' with no nested schema; get_tool_parameters' 'tool' field is typed as 'object' with no structure.
Descriptions are too brief (10-50 chars vs. 50-200 char baseline). 'List all available tools from the MCP server' (50 chars) lacks context on when to call it or what structure it returns.
Output schemas are completely undocumented. No tool declares what fields it returns, what types they are, or how to chain results to downstream calls.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
No error handling guidance. If call_tool fails (invalid tool name, bad parameters, timeout), there is no description of error format or recovery steps.
Parameter 'parameters' in call_tool is untyped and undescribed. Should specify: is it a JSON object? What keys/values are valid? How does the tool validate them?