生态洞察 · 2026-W44

重点关注 · 6 个异常信号

中本周新增 3986，近 3 周均值 1088——突增，可能有批量导入/刷库

中账号 uckkk 有 1286 个插件、均分仅 56——疑似批量刷库/模板复制

中账号 CAI-MH 有 9 个插件、均分仅 57、1 个无 README——疑似批量刷库/模板复制

低最大账号 uckkk 独占 1286 个插件（占全生态 7%）、均分 56——模板批量号拉高规模数字，解读增长时需剔除水分

低10 个插件仓库版本领先 npm 发布版本——安装侧拿到的是旧版

高dsh 官方本周发布 3 个版本，其中 1 个含 breaking 变更（dsh-v0.2.2-alpha.2）——插件作者需评估适配

DSH 生态洞察 · 2026-W44（2026/10/26～2026/11/01）

由 DSH Insights 管线 + AI 生成 · 数据快照 2026-11-02 · 启发式评估，非安全审计

1. 核心结论

免费模型入口成最大流量池：Ebony-Vinyl/dsh-our-free-model 以 55,755 stars 和 12,073 周下载占据绝对头部，说明用户首要诉求是零门槛接入模型，但该插件集中在单一账号，存在单点风险。
高下载不等于高质量：ranxianglei/billion-context 周下载 788,347 为全生态最高，但 grade 仅 B（86），且明星插件 dsh-market/dsh-market 存在 npm 版本漂移——安装侧拿不到最新代码。
批量账号污染规模指标：uckkk 账号持有 1,286 个插件（均分 56），占全生态 7%，让新增与总数数字失真，真实健康度应剔除该账号后评估。
队列质量在提升：最近四周新插件队列平均分从 77.3 升至 78.3（本周样本仅 274 个，需谨慎），新增者整体工程素养高于存量均值 74.1。
官方迭代进入快节奏：本周 dsh 发布 3 个版本，其中 dsh-v0.2.2-alpha.2 含 breaking 变更，插件作者若不跟进可能直接失去运行时兼容。

2. 头部插件深潜

Ebony-Vinyl/dsh-our-free-model：以“无需登录/API Key 即可用 DeepSeek V4.1 Flash、Kimi K3”为卖点，55,755 stars、A级93分、周下载12,073。赢在直击用户痛点，但集中在单一账号且依赖第三方模型路由，存在服务中断风险。
dsh-market/dsh-market：官方市场插件，10,113 stars、A级90分，周下载361,739，是发现与安装的核心通道；但 npm 版本落后仓库（npm 1.73.8 vs 仓库 1.43.0），用户拿不到最新市场元数据，需尽快修复。
ranxianglei/billion-context：通用上下文压缩代理，周下载788,347为最高，但 grade B（86），drops 中有 npm.version-drift 与 manifest.no-client-export；高采用+中等评分说明解决真实痛点，但工程细节拖后腿。
sunyuhuirong/deepseek-account：仅 1 star、无描述、无分类，周下载却达559,082；这是一个“隐形工具”，大概率靠 npm 包名或默认引用获得安装，用户在不了解质量的情况下大规模使用，需警惕。

头部格局表明：生态当前由模型接入与基础设施插件主导，但流量分配并不均匀——一个工具可以没有 star 却获得巨量安装，发现与信任机制存在断层。

3. 作者格局

头部集中 vs 长尾：作者总数 11,496，多插件作者 2,884；Ebony-Vinyl 4 个插件共 56,375 stars，占全生态 star 约 21%（按 dshStars 268,416 计算），一个账号牵动整个生态的注意力。
高产低质批量号：xby-skill 175 个插件、均分 62.2、仅 2 stars；uckkk 1,286 个插件、均分 56；这类账号为凑数而生成模板插件，严重影响排行榜可信度。
高质量小作者被埋没：Fish-under-sea 4 个插件均分 96.5 但仅 20 stars；he0119 3 个插件均分 95、3 stars；他们代表社区自发的高水平贡献，却难以获得曝光。
连接者作用有限：connectors 列表中 OMSociety（5 插件、57,843 stars）和 Ebony-Vinyl 等更多的是自身 star 高，协作图关键节点并未催生跨团队项目。
判断：这个生态目前是“流量驱动”与“社区驱动”并存——病毒式插件带来大量新人，但批量号稀释了作者榜的真实性，真正靠质量说话的作者仍需官方策展或推荐机制扶持。

4. 质量 × 采用度交叉发现

被低估宝藏：
- SenmuuuuW/dsh-whale-report：S 级 100 分、0 stars、周下载 257；从会话事件自动生成日报/周报/年报，是监控场景的杀手级补充，极可能因为缺少 star 而被搜索排序压制。
- dsh-plugin-lab/dsh-workbuddy-bridge：S 级 100 分、28 stars、周下载 2,890；零配置接入 WorkBuddy 模型，适合混合工具链用户。
低质高用风险：
- LuniteGlaze/dsh-plugin-catalog：C 级 64 分、1 star、周下载 136,941；无描述、30 天无提交，大量用户在一个不透明且可能停更的目录插件上依赖。
- Creakono/dsh-cost-meter：C 级 64 分、1 star、周下载 44,084；成本计算若出错会直接误导用户对 API 费用的判断，缺乏维护与测试是硬伤。
so-what：用户习惯于按下载量选择插件，但本周数据多次出现“高下载低分”，官方应在市场页强制展示健康分与维护状态。

5. 生态主题地图

主题聚类（基于 LLM 标签，覆盖率仅 1,823/19,503，结论有限）：session、search、ui、memory、skills、web、agent 是高频标签，反映生态正从“聊天增强”走向“智能体工作台”。
代表插件映射：模型接入与路由——Ebony-Vinyl/dsh-our-free-model、yjh051108/dsh-routing-suite；监控与用量——DeepSeek-Balance-Whale-Widget、T-Auto/dsh-ops、billion-context；视觉与图像——liustack/modlens、HellowVirgil/dsh-antv-ava；效率与桌面宠物——dsh-pet 系列（levi52/dsh-pet、ALKAERR/dsh-pet 等）出现多个 fork，已显拥挤。
拥挤方向：其它 类目 6,613 个插件、均分 71.3，是低质量插件的收纳箱；主题/视觉美化 771 个、均分 73.9，同质化严重；工具/效率 951 个、均分 74.6，中位数创新不足。
空白机会：官方 scenario 面板 8 个场景（如 file-explore、memory-knowledge）各仅收录 6 个插件，远低于生态总量；作者若针对官方场景精做高质量插件，被收录概率与曝光将显著提升。

主题地图显示空位在官方场景精选与中文文档，插件作者应聚焦未被满足的工作流而非继续堆叠宠物和主题。

6. 生态健康度评估

维度	本期数值	解读
规模	authoritative 19,503；active30 9,024（46.3%）	总量接近 2 万，但仅 46.3% 在近 30 天有活动，一半以上疑似休眠或一次性提交。
增长	weeklyDiff added 3,986（基线 2026-10-15，18天累计，非标准周）	相对基线的累计增长，非本周新增；若按近 3 周均值 1,088 计算，增长脉冲显著，需甄别批量导入。
质量分布	S 186、A 802、B 9,107、C 6,975、D 2,433；平均分 74.1	B 和 C 占 82.5%，可用的中间层庞大；S+A 合计仅 5.1%，头部精品稀缺。
活跃度	active7 3,062（15.7%）；active30 9,024（46.3%）	近 7 天活跃率仅 15.7%，大量插件为一次性提交后不再维护，长尾依赖风险高。
npm 发布率	publish 44%（8,574 发布 / 10,929 未发布 / 4,745 stale）	超过一半插件未发布到 npm，安装侧依赖 Git 或复制包，版本管理与安全难以保障。
文档覆盖	README 19,451（99.7%）；zh 覆盖率 48.3%；双语 both 9,414	README 基本普及，但中文文档不足一半，与 DSH 中文社区主流不符，影响非英语用户上手。

Cohort 趋势：新插件队列平均分从 2026-08-24 的 73.7 升至 2026-10-26 的 78.3，最近四周稳定在 77+，说明新增作者的工程意识在改善；但本周样本仅 274 个（数据截断），且批量号会拉低真实均值，仍需持续观察后续完整周数据。

7. 异常与应对

growth-anomaly（severity: mid）

可能原因（假设）：某自动化工具或模板库在 2026-10-15 后批量导入了一批插件，或官方审核通道积压后集中放行。
影响面：新增数字虚高，top-new 榜单被刷，真实社区新提交被淹没；后端存储与索引压力增大。
应对建议：对 2026-10-15 至 10-26 期间新增的 3,986 个插件做来源 IP/账号聚类，标记批量号并冻结刷库账号；官方可临时提高上架门槛（如要求 README 与唯一描述）。

owner-cluster（severity: mid）— `uckkk`

可能原因（假设）：uckkk 使用脚本批量生成同名前缀插件，每一件只改一个参数或模板文本，目的是占坑或提升账号权重。
影响面：1,286 个低质插件平均分仅 56，拉低全生态平均分与作者榜可信度；用户搜索时被这些插件的名称污染。
应对建议：官方封禁或降权该账号，并对 56 分以下、无下载或无 star 的批量插件做下架处理，同时在榜单中增加“去批量号”筛选。

owner-cluster（severity: mid）— `CAI-MH`

可能原因（假设）：与 uckkk 类似，CAI-MH 的 9 个插件可能来自同一模板复制，其中 1 个甚至缺失 README，说明未经过人工审核。
影响面：虽然绝对数量少，但暴露出发布流程中模板复制的漏洞，若不阻断，未来可能出现更多小批量号。
应对建议：对 9 个插件逐一人工复核，删除无 README 项；增加“相同 README 指纹”检测，防止模板克隆。

dominant-owner（severity: low）— `uckkk`

可能原因（假设）：同一原因，批量号主导了规模指标，但 star 与下载未同步增长，说明是纯数量填充。
影响面：官方在对外发布“插件总数”“增速”等指标时会被质疑水分，生态健康评估需剔除该账号后重算。
应对建议：在 analytics 后台增加“排除 top 批量号”的视图，并在周报中同时公布剔除后的真实规模与增速。

npm-stale（severity: low）

可能原因（假设）：插件作者未配置 CI 自动发布，或本地仓库版本号更新后忘记发布 npm，导致 npm 上仍是旧包。
影响面：以 dsh-market 为例，npm 版本 1.73.8 领先仓库 1.43.0（有抢注/错配可能），用户安装到的可能不是作者当前维护的版本；10 个插件受影响。
应对建议：为 top 下载插件建立自动 npm publish 流程；市场页对 repo 与 npm 版本不一致的插件显示“版本漂移”警告，并提示用户使用 Git 安装。

breaking-release（severity: high）— `dsh-v0.2.2-alpha.2`

可能原因（假设）：官方为修复关键缺陷或重构内部 API 而引入了不向后兼容的变更，并选择在 alpha 通道快速验证。
影响面：所有调用旧内部接口的插件可能在最新运行时上崩溃；若未明确标注，用户升级后可能遇到插件全部失效。
应对建议：官方发布 breaking 版本时应同步提供迁移清单；插件作者立即用 alpha 版本跑兼容测试，并在 package.json 中声明兼容 dsh 版本范围；用户在 alpha 阶段暂缓生产环境升级。

8. 下一阶段重点与建议

插件作者

检查 npm-stale 列表中的版本漂移，配置 GitHub Actions 自动发布 npm。
优先补中文文档（zh 覆盖率仅 48.3%），尤其面向中文社区的功能。
远离 uckkk 式模板复制，专注官方 scenario 中空白场景（如 memory-knowledge）。
用 dsh-v0.2.2-alpha.2 做兼容性测试，并在插件元数据中标注支持版本。

生态用户

不只看下载量，优先选择 grade A/S 且维护活跃的插件，如 quadrant.gems 中的 SenmuuuuW/dsh-whale-report。
对高下载但 grade C 的插件（如 LuniteGlaze/dsh-plugin-catalog）保持警惕，查看其最近提交时间与描述。
在生产环境暂缓使用官方 breaking 版本，待插件作者适配后再升级。
关注官方市场中的“版本漂移”与“维护停滞”标签，避免安装陈旧包。

DSH 官方

封禁或降权 uckkk 等批量号，清理模板插件，发布剔除水分后的规模报告。
为 npm 版本漂移提供自动检测与告警，并在市场页强制展示健康分。
扩展现有 8 个官方场景的收录数量，从 6 个/场景增加到至少 30 个，给高质量作者更多分发位。
发布 breaking 变更前提供 migration guide，并在文档站首页设置醒目升级公告。

9. 官方动态解读

dsh-v0.2.1-rc.5（2026-10-30，breaking false）与 dsh-v0.2.2-alpha.2（breaking true）同周发布，显示官方并行维护稳定与激进双通道；插件作者需紧盯 alpha 变更。
dshStars 268,416 且 dshPushed 2026-10-30，说明核心仓库活跃；周边官方项目 DeepSeek-V3（104,709 stars）与 R1（91,993 stars）仍为生态导流，插件市场仍有红利。
官方 awesome-deepseek-integration（39,672 stars）等聚合项目未与插件市场数据打通，市场发现插件的能力弱于外部 star 效应，需加强内部策展。

DSH Ecosystem Insights · 2026-W44（2026/10/26～2026/11/01）

Generated by the DSH Insights pipeline + AI · snapshot 2026-11-02 · heuristic evaluation, not a security audit

1. Core Conclusions

Free model entry becomes biggest traffic pool: Ebony-Vinyl/dsh-our-free-model dominates with 55,755 stars and 12,073 weekly downloads, showing users' first need is zero-barrier model access, but concentration in a single account introduces single-point risk.
High downloads do not equal high quality: ranxianglei/billion-context has the ecosystem's highest weekly downloads at 788,347 but grade B (86); star plugin dsh-market/dsh-market suffers npm version drift—users do not get the latest code.
Batch accounts pollute scale metrics: account uckkk holds 1,286 plugins (avg score 56), 7% of the ecosystem, distorting totals and growth; true health should be evaluated after removing it.
Cohort quality is rising: average score of recent four weekly plugin cohorts rose from 77.3 to 78.3 (this week's sample only 274, treat with caution), indicating new contributors have better engineering practices than the stock average of 74.1.
Official iteration accelerates: dsh shipped 3 releases this week, among which dsh-v0.2.2-alpha.2 contains breaking changes; plugin authors who fail to follow may lose runtime compatibility.

2. Deep Dive into Top Plugins

Ebony-Vinyl/dsh-our-free-model: Sells “no login/API Key required to use DeepSeek V4.1 Flash, Kimi K3”; 55,755 stars, grade A score 93, weekly downloads 12,073. Wins by hitting user pain points, but concentrated in one account and relies on third-party model routing—service interruption risk.
dsh-market/dsh-market: Official marketplace plugin, 10,113 stars, grade A score 90, weekly downloads 361,739; core channel for discovery and installation. But npm version lags repo (npm 1.73.8 vs repo 1.43.0), users don't get latest market metadata—needs urgent fix.
ranxianglei/billion-context: Universal context-compression proxy, highest weekly downloads 788,347, but grade B (86) with npm.version-drift and manifest.no-client-export. High adoption + moderate score means it solves real pain, but engineering details lag.
sunyuhuirong/deepseek-account: Only 1 star, no description, no category, yet weekly downloads 559,082; an “invisible tool” likely installed via npm package name or default reference. Users mass-adopt without quality awareness—must be vigilant.

Top pattern: ecosystem currently driven by model access and infrastructure plugins, but traffic is uneven—a tool can have no stars yet massive installs; discovery and trust mechanisms are broken.

3. Author Landscape

Head concentration vs long tail: 11,496 authors, 2,884 multi-plugin authors; Ebony-Vinyl has 4 plugins with 56,375 stars, about 21% of total ecosystem stars (based on dshStars 268,416)—one account sways the entire ecosystem's attention.
High-volume low-quality batch accounts: xby-skill 175 plugins, avg score 62.2, only 2 stars; uckkk 1,286 plugins, avg 56; these accounts generate template plugins to inflate numbers, severely harming leaderboard credibility.
High-quality small authors are buried: Fish-under-sea 4 plugins avg score 96.5 but only 20 stars; he0119 3 plugins avg 95, 3 stars; they represent high-level community contributions but receive little exposure.
Connectors have limited effect: connectors list shows OMSociety (5 plugins, 57,843 stars) and Ebony-Vinyl mostly rely on their own star power; collaboration graph nodes have not spawned cross-team projects.
Judgment: this ecosystem is both “traffic-driven” and “community-driven”—viral plugins bring new users, but batch accounts dilute author credibility; quality-first authors still need official curation or recommendation support.

4. Quality × Adoption Cross Findings

Undervalued gems:
- SenmuuuuW/dsh-whale-report: S grade 100, 0 stars, weekly downloads 257; auto-generates daily/weekly/monthly reports from session events—a killer supplement for monitoring, likely suppressed by search ranking due to zero stars.
- dsh-plugin-lab/dsh-workbuddy-bridge: S grade 100, 28 stars, weekly downloads 2,890; zero-config integration of WorkBuddy models, suitable for hybrid toolchain users.
Low-quality high-adoption risks:
- LuniteGlaze/dsh-plugin-catalog: C grade 64, 1 star, weekly downloads 136,941; no description, 30 days without commit; many users depend on an opaque and possibly discontinued catalog plugin.
- Creakono/dsh-cost-meter: C grade 64, 1 star, weekly downloads 44,084; if cost calculation errors, it directly misleads users on API expenses; lack of maintenance and tests is a hard flaw.
So-what: users habitually choose plugins by download volume, but this week's data repeatedly shows “high downloads low score”; official market should display health score and maintenance status prominently.

5. Ecosystem Theme Map

Theme clusters (based on LLM tags, coverage only 1,823/19,503, conclusions limited): session, search, ui, memory, skills, web, agent are high-frequency tags, reflecting shift from “chat enhancement” to “agent workbench”.
Representative plugin mapping: model access/routing—Ebony-Vinyl/dsh-our-free-model, yjh051108/dsh-routing-suite; monitoring/usage—DeepSeek-Balance-Whale-Widget, T-Auto/dsh-ops, billion-context; vision/image—liustack/modlens, HellowVirgil/dsh-antv-ava; efficiency/desktop pet—dsh-pet series (levi52/dsh-pet, ALKAERR/dsh-pet) has multiple forks, already crowded.
Crowded directions: Other category has 6,613 plugins avg score 71.3, a dumping ground for low-quality plugins; Theme/Visual 771 plugins avg 73.9, high homogeneity; Tools/Efficiency 951 plugins avg 74.6, median innovation insufficient.
Blank opportunities: official scenario panel has 8 scenarios (e.g., file-explore, memory-knowledge) each only 6 plugins, far below ecosystem total; authors focusing on high-quality plugins for official scenarios will gain significant collection probability and exposure.

Theme map shows gaps in official scenario curation and Chinese documentation; authors should focus on unmet workflows rather than piling up pets and themes.

6. Ecosystem Health Assessment

Dimension	Current Value	Interpretation
Scale	authoritative 19,503; active30 9,024 (46.3%)	Total near 20k, but only 46.3% active in last 30 days—over half suspected dormant or one-time submissions.
Growth	weeklyDiff added 3,986 (baseline 2026-10-15, 18-day cumulative, not standard week)	Cumulative relative to baseline, not this week's new; compared to 3-week mean 1,088, growth pulse significant—must screen batch imports.
Quality distribution	S 186, A 802, B 9,107, C 6,975, D 2,433; avg score 74.1	B and C account for 82.5%, large middle tier; S+A only 5.1%, top-quality scarce.
Activity	active7 3,062 (15.7%); active30 9,024 (46.3%)	Only 15.7% active in last 7 days; many one-time submissions never maintained, long-tail dependency risk high.
npm publish rate	publish 44% (8,574 published / 10,929 unpublished / 4,745 stale)	Over half not published to npm; install side relies on Git or copied packages, version management and security difficult to guarantee.
Documentation coverage	README 19,451 (99.7%); zh coverage 48.3%; bilingual both 9,414	README nearly universal, but Chinese docs less than half, mismatched with DSH Chinese community mainstream, hurting non-English users.

Cohort trend: new plugin cohort average score rose from 73.7 on 2026-08-24 to 78.3 on 2026-10-26, stable above 77 in recent four weeks—new authors' engineering awareness improving; but this week's sample only 274 (data truncated), and batch accounts may lower true mean; continue observing full weeks.

7. Anomalies and Responses

growth-anomaly (severity: mid)

Possible cause (hypothesis): Some automation tool or template library batch-imported plugins after 2026-10-15, or official review channel backlogged and released centrally.
Impact: New additions inflated, top-new list gamed, real community submissions submerged; backend storage and indexing pressure increased.
Response: Cluster source IP/accounts for the 3,986 plugins added between 2026-10-15 and 10-26, flag batch accounts and freeze farming accounts; official may temporarily raise listing threshold (e.g., require README and unique description).

owner-cluster (severity: mid) — `uckkk`

Possible cause (hypothesis): uckkk uses scripts to batch generate same-prefix plugins, only changing one parameter or template text to occupy slots or boost account weight.
Impact: 1,286 low-quality plugins with avg 56 drag down ecosystem average and leaderboard credibility; users' search results polluted by these names.
Response: Official should ban or downrank the account, remove batch plugins under score 56 with no downloads or stars, and add a “remove batch accounts” filter in leaderboards.

owner-cluster (severity: mid) — `CAI-MH`

Possible cause (hypothesis): Similar to uckkk, CAI-MH's 9 plugins may come from the same template copy, with one missing README, indicating no manual review.
Impact: Though small in absolute count, it exposes template-copy vulnerabilities in publishing; if not blocked, more small batch accounts may appear.
Response: Manually review all 9 plugins, delete the one without README; add “identical README fingerprint” detection to prevent template cloning.

dominant-owner (severity: low) — `uckkk`

Possible cause (hypothesis): Same cause—batch account dominates scale metrics, but stars and downloads have not grown correspondingly, indicating pure quantity filling.
Impact: Official published metrics like “total plugins” and “growth rate” will be questioned for water; ecosystem health assessment must recalculate after excluding this account.
Response: Add a “exclude top batch accounts” view in analytics backend, and publish real scale/growth after exclusion in weekly reports.

npm-stale (severity: low)

Possible cause (hypothesis): Plugin authors did not configure CI auto-publish, or forgot to publish after updating local repo version, leaving old package on npm.
Impact: Taking dsh-market as example, npm version 1.73.8 is ahead of repo 1.43.0 (possible squatting/mismatch); users may install a version not maintained by author; 10 plugins affected.
Response: Set up automatic npm publish workflow for top-download plugins; market page should display “version drift” warning for repo/npm mismatch and suggest Git installation.

breaking-release (severity: high) — `dsh-v0.2.2-alpha.2`

Possible cause (hypothesis): Official introduced non-backward-compatible changes to fix critical defects or refactor internal APIs, choosing fast validation in alpha channel.
Impact: All plugins calling old internal interfaces may crash on latest runtime; if not clearly marked, users may face all plugins failing after upgrade.
Response: Official should provide migration checklist with breaking release; plugin authors should immediately run compatibility tests with alpha and declare supported dsh version range in package.json; users should postpone production upgrade during alpha.

8. Next Steps & Recommendations

Plugin authors

Check version drift in npm-stale list and configure GitHub Actions auto-publish to npm.
Prioritize Chinese documentation (zh coverage only 48.3%), especially for Chinese community features.
Stay away from uckkk-style template copying; focus on blank official scenarios like memory-knowledge.
Test compatibility with dsh-v0.2.2-alpha.2 and declare supported version in plugin metadata.

Ecosystem users

Don't only look at download counts; prefer grade A/S and actively maintained plugins, like SenmuuuuW/dsh-whale-report from quadrant.gems.
Be cautious with high-download grade C plugins (e.g., LuniteGlaze/dsh-plugin-catalog); check recent commit time and description.
Hold off production upgrades to official breaking versions until plugin authors adapt.
Pay attention to “version drift” and “maintenance stale” labels in official market, avoid installing outdated packages.

DSH official

Ban or downrank batch accounts like uckkk, clean template plugins, and publish water-removed scale reports.
Provide automatic detection and alerts for npm version drift, and force display of health score on market page.
Expand current 8 official scenarios from 6 to at least 30 plugins each, giving high-quality authors more distribution slots.
Publish migration guide before breaking changes, and place prominent upgrade notice on docs homepage.

9. Official Dynamics Interpretation

dsh-v0.2.1-rc.5 (2026-10-30, breaking false) and dsh-v0.2.2-alpha.2 (breaking true) released same week, showing official runs stable and aggressive channels in parallel; plugin authors must watch alpha changes.
dshStars 268,416 and dshPushed 2026-10-30 indicate active core repo; surrounding official projects DeepSeek-V3 (104,709 stars) and R1 (91,993 stars) still funnel traffic, so plugin market still has dividend.
Official aggregation project awesome-deepseek-integration (39,672 stars) is not connected with plugin market data; market's discovery ability is weaker than external star effect—needs stronger internal curation.