Loading ToolBench
MCP server for the next_best_prompt 'next-best-prompts' end-of-turn rule — end a substantive turn with 2-4 ranked, copy-paste-ready next moves, and skip when there's none. Pure guidance: no secrets, no network, no side-effecting tools.
Single tool with empty input schema ({}). Description is present but generic (89 chars, below 100-char minimum for clarity). No parameter descriptions exist because there are no parameters. Tool is read-only guidance with no side effects, but lacks the structured output documentation needed for LLM integration. The tool name 'get_next_best_prompts_rule' follows verb_noun convention (+), but the schema and output structure are not documented in the source code provided.
Returns the next-best-prompts end-of-turn rule text. The rule defines how to surface 2-4 ranked, copy-paste-ready next moves at the end of a substantive response.
Input schema is empty ({}). No parameters are defined or documented. Per HARD SCORING RULES, schema score must be 0 when no schema is visible.
Output schema is not documented in source code. LLMs cannot plan downstream actions or extract fields without knowing the response structure (e.g., is it a string, object, array?).
Tool description is 89 characters, below the 100-char minimum for clarity. It lacks WHEN to use this tool and WHAT the LLM should do with the returned rule text.
No error handling guidance. If the rule cannot be retrieved or is malformed, the tool provides no recovery path for the LLM.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 40 | 2026-07-28+ | v2 |