Loading ToolBench
A Next.js website for mcpz, a CLI tool for managing Model Context Protocol (MCP) tools and optimizing token usage. Includes documentation, blog, pricing, and a discover platform for MCP resources.
Scoring was not performed
Missing documented output schemas. Neither tool declares what fields callers should expect in the response. This forces LLMs to guess at response structure and plan downstream tool calls blindly.
Parameter descriptions are minimal or absent. 'platform' and 'search' for discover_repositories lack guidance on format, valid values, and behavior. No description states which platforms are supported or how search filtering works. 'repoId' and 'platform' for create_mcp_from_slop have no format or validation guidance.
Tool descriptions lack actionability. They state WHAT the tools do but not WHEN to use them or what to do if they fail. 'discover_repositories' description mentions sorting by type and activity but doesn't explain when to call it vs. alternatives. 'create_mcp_from_slop' description doesn't explain prerequisites (does repoId need to exist first?) or what 'validation' means.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 18 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 32 | - | v1 |
No error handling or recovery guidance. If discover_repositories receives an invalid platform, no error description guides the LLM to retry with valid values. If create_mcp_from_slop fails validation, the LLM gets nothing to act on.
create_mcp_from_slop performs state mutation (creates MCP template) without idempotency guarantees or confirmation. No description warns that this is a write operation with potential side effects. Risk marked as WRITE but no retry-safety or confirmation pattern documented.
discover_repositories lacks pagination pattern. If results exceed context limits, no limit/offset/page parameters are documented. Returning all repositories for a platform could blow token budgets.
Parameter 'platform' does not declare an enum. Valid platforms (valtown, codesandbox, npm, pip, github) are mentioned in the description text, but LLMs cannot parse text constraints reliably. Enum declaration in the schema would be self-documenting and prevent hallucinated values.