Loading ToolBench
MCP server for detecting and converting Mainland Chinese terminology to Traditional Chinese (Taiwan) usage. Provides dictionary lookup, fuzzy search, custom term management, and automatic dictionary updates from multiple sources.
Server defines 8 tools with complete JSON schemas and descriptions in Traditional Chinese. All tools have explicit input schemas with typed properties and required fields. Naming follows verb_noun convention (convert_text, lookup_term, search_terms, list_categories, add_term, remove_term, update_dictionary, get_stats). Descriptions are present (avg ~80 chars) but lack LLM-optimized guidance on WHEN to use each tool vs alternatives, and do not explicitly state side effects for write/destructive operations. Parameter descriptions are minimal (5-20 chars) and lack format/constraint details. Output schemas are not documented, LLMs cannot predict response structure. Error handling returns text messages but lacks recovery guidance or error classification. No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels (READ_ONLY, WRITE, DESTRUCTIVE).
新增自訂的大陸→台灣用語對照。會存入 ~/.che-tw-terms-mcp/custom.json。
偵測並替換中國大陸用語為台灣繁體中文慣用語。回傳轉換後的文字及變更清單。
顯示詞庫統計資訊，包括詞條數量、來源分布、分類分布、最後更新時間。
列出所有詞彙分類及各分類的詞條數量。
查詢一個詞的大陸↔台灣對照。可輸入大陸或台灣用語。
移除自訂的用語對照（僅限 custom.json 中的詞條）。
模糊搜尋詞庫，支援大陸或台灣用語。
Output schemas not documented. LLMs cannot predict response structure (e.g., what fields does convert_text return? Is it {text, changes} or {original, converted, diff}?). Forces agents to guess and parse unstructured output.
Tool annotations missing. Risk levels (READ_ONLY, WRITE, DESTRUCTIVE) are documented in metadata but not exposed via MCP tool annotations (readOnlyHint, destructiveHint, idempotentHint). Agents cannot determine which tools are safe to retry or have side effects.
Parameter descriptions are minimal (5-20 chars) and lack format/constraint guidance. E.g., 'limit' param in search_terms has no min/max bounds; 'source' param in update_dictionary lacks enum constraint documentation in description (only in code comment).
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 65 | 2026-07-28+ | v2 |
從線上來源更新詞庫。支援 g0v 教育部資料和 Wikipedia 兩岸用語列表。
Error handling returns plain text messages without recovery guidance or error classification. E.g., 'Error: term is required' tells LLM nothing about whether to retry, ask user, or abandon. No distinction between retryable (network) vs user-fixable (invalid input) vs fatal errors.
Tool descriptions lack WHEN-to-use guidance. E.g., lookup_term vs search_terms distinction is unclear, when should agent call one vs the other? No guidance on dependencies (e.g., 'Call list_categories first to see available categories before add_term').