Loading ToolBench
番茄雷达 — 番茄小说榜单 + 红果短剧榜 / 扫榜选书工作台 / MCP Server（fanqie novel rank data, hongguo drama rank, webnovel trend analysis, imitation book scoring）。零配置、零第三方运行时依赖（纯标准库，含 urllib 采集器），内置当天公开榜单真实快照装完即用，附极速一键扫榜与一键安装 Agent Skill。
fanqie-radar provides 10 tools with complete input schemas and descriptions in Chinese. Naming follows verb_noun patterns (fanqie_ranks, fanqie_find, fanqie_trend). Descriptions are substantive (100-200 chars), explaining WHAT and WHEN. However, parameter descriptions lack depth, most are 1-2 lines without format/range guidance. Output schemas are undocumented; responses are inferred from code (dict returns with 'count', 'items', 'breakdown'). Error handling is minimal, no recovery guidance or actionable error messages. Security is sound (read-only operations, no secrets in params, safe field projection). Tool composition is clean (single responsibility), but some tools like fanqie_imitation_score are complex and could benefit from clearer breakdown documentation.
跨全部快照按书名/作者模糊查找，返回 book_id 与书目信息。
题材热度聚合（按在读总量排序），识别当前吸量赛道。
书名热词频次统计（基于套路词表），快速看当前书名流行词。
【核心】仿写选书六维评分（完结度/体量适配/单本热度/题材吸量/套路密度/金手指清晰 + 稳定性/跨榜/巅峰加成）。也可切男频连载母本评分。用于在读爆款里筛出'最适合仿写'的书，并给出每本的分数拆解。
读取最新番茄榜单（按频道/榜型/品类）。返回书名、作者、在读、状态、章节数、book_id——不含简介全文与版权图。
数据概览：快照数量、书目总数、book_id 覆盖率、被探测到的数据目录。
多日趋势差分：对比同一榜单最新两份快照，给出新上榜/掉榜/排名变化/在读增长。需 data/ 下存在至少两份同榜快照。
Output schemas undocumented. Code infers responses return {count, items, breakdown, ...} but no formal schema is declared. LLMs cannot plan downstream operations without knowing response structure.
Parameter descriptions lack actionable constraints. 'top' is described as '返回条数' (return count) but no min/max bounds stated. 'channel' enum is clear, but 'category' accepts free-form strings with only a vague example ('豪门总裁'). LLMs cannot validate inputs without explicit ranges.
No error handling or recovery guidance. Tools return bare dicts on success but no documented error cases. If data/ directory is missing or snapshots are insufficient (e.g., fanqie_trend needs 2+ snapshots), error messages are not shown in code. LLMs cannot self-correct.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 59 | 2026-07-28+ | v2 |
跨全部红果短剧快照按剧名模糊查找，返回 series_id 与热度。
读取最新红果短剧榜（热播/AI/漫画/真人四榜）。返回剧名、series_id、热度。数据来自本地快照（fanqie_drama_ranks_*.json），可用 `python -m fanqie_index.drama --scrape` 采集。
红果短剧数据概览：快照数量、剧集总数（series_id 去重）、各榜条数、最新快照日期。
fanqie_imitation_score is complex (6+ scoring dimensions, breakdown fields, trope/gf hits) but breakdown structure is not formally documented. Code shows it returns {s_done, s_size, s_reads, s_heat, s_trope, s_gf, s_stable, trope_hits, gf_hits} but no schema or field descriptions provided.
fanqie_stats and hongguo_stats have empty input schemas (no properties). Descriptions are minimal (15-20 chars). These discovery tools should explain what data they reveal and when to call them first.