Loading ToolBench
MCP server with tools for arithmetic operations and Twitter/X posting
Both tools have critical definition gaps. addTwoNumbers and createPost lack parameter descriptions entirely (null values in schema). Tool descriptions are present but minimal (10-34 chars, below the 50-200 char LLM-optimized baseline). No output schemas documented. createPost exposes a write operation without error handling, recovery guidance, or confirmation pattern. Parameter naming lacks type suffixes (e.g., 'status' is ambiguous). No input validation or constraint documentation visible.
Add two numbers
Create a post on X formally known as Twitter
All parameter descriptions are null. LLMs cannot infer parameter meaning from names alone. 'a' and 'b' in addTwoNumbers are ambiguous; 'status' in createPost could mean HTTP status, project status, or boolean flag.
Tool descriptions are too short (10-34 chars vs 50-200 char baseline). 'Add two numbers' and 'Create a post on X formally known as Twitter' lack context for LLM selection, prerequisites, or when to use each tool.
createPost is a destructive write operation with no error handling, recovery guidance, confirmation pattern, or dry-run support. Agents cannot know if the call succeeded, what to do on failure, or how to undo accidental posts.
No output schemas documented. LLMs cannot plan downstream tool calls or extract required fields. Responses are hardcoded text strings without structured field definitions.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Twitter API credentials (appKey, appSecret, accessToken, accessSecret) are injected via environment variables, which is correct. However, no audit logging, permission gates, or rate limiting visible. Agents can post unlimited tweets without traceability.