Loading ToolBench
AI investigation memory — single-binary MCP server for managing investigation notes, topics, and policy rules with SQLite indexing and Claude Code integration
Single tool 'echo' with minimal definition. Tool has a valid input schema (object with required 'text' string property) but description is extremely terse (16 chars: 'echo input back'). No parameter descriptions provided, the 'text' parameter lacks any guidance on format, constraints, or purpose. No output schema documented. No error handling guidance. Tool is read-only but lacks any annotation or documentation of this. Naming is generic and does not follow verb_noun pattern (should be 'echo_text' or similar). Overall definition quality is well below production baseline.
echo input back
Tool description is 16 characters ('echo input back'), below 20-char minimum. Provides no context on when to use, what it returns, or parameter semantics. LLM cannot determine tool purpose from this description.
Parameter 'text' has no description. LLM cannot infer what format, constraints, or semantics apply. Rubric requires every parameter to have a non-empty description explaining what it controls.
No output schema documented. LLM cannot predict what fields or structure the tool returns, making it impossible to chain with other tools or extract specific data.
Tool name 'echo' is generic and does not follow verb_noun pattern. Should be 'echo_text' or 'echo_message' to clarify the action and object. Generic names reduce LLM's ability to disambiguate when many tools are available.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 38 | 2026-07-28+ | v2 |
No error handling guidance. Tool implementation returns ErrorResult for unknown tool names and bad args, but descriptions do not explain what errors are possible or how to recover.