Loading ToolBench
A fault-injection playground for debugging MCP clients & servers. Exposes 40+ deliberately broken MCP endpoints, each reproducing a specific real-world failure mode.
chaos-mcp is a fault-injection test server, not a production tool suite. The 3 tools (echo, add, current_time) have minimal descriptions (10-34 chars, well below the 50-200 char LLM-optimized baseline). Schemas are present but sparse: 'echo' and 'add' lack parameter descriptions; 'current_time' timezone param is undescribed. No error handling guidance, no output schemas documented, no tool annotations. The server's purpose is to corrupt MCP deliberately, so quality gaps are intentional, but by production standards, this scores as poor.
Add two numbers and return the sum.
Return the current server time in RFC3339.
Echo back the provided message.
Tool descriptions are critically short (10-34 chars). LLM-optimized descriptions should be 50-200 chars, explaining WHAT the tool does, WHEN to use it, and any prerequisites.
Parameter descriptions are missing or absent. 'echo' message param, 'add' a/b params, and 'current_time' timezone param lack descriptions. LLMs cannot infer parameter meaning from names alone.
No output schemas documented. LLMs need to know what fields to expect so they can plan downstream calls and extract data correctly.
No error handling guidance. Error responses should tell the LLM what to do next (retryable, user-fixable, or fatal) and include the invalid value and constraint violated.
Inferred effective spec: 2026-07-28+.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | 2026-07-28+ | v2 |
No tool annotations (readOnlyHint, destructiveHint, idempotentHint). All three tools are marked READ_ONLY in metadata but lack formal annotation in schema.