生态洞察 · 2026-W43
重点关注 · 4 个异常信号
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 8%）、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
DSH 生态洞察 · 2026-W43（2026/10/19～2026/10/25）
由 DSH Insights 管线 + AI 生成 · 数据快照 2026-10-26 · 启发式评估，非安全审计
1. 核心结论
- 生态规模被灌水污染：uckkk 单账号 1286 个插件、均分 56，占全库 8%，有效供给远低于 17011。
- 采用度与质量背离：下载榜 Top10 中 dsh-pet（C,周 46041）、schedule（C,周 29926）、dsh-cost-meter（C,周 24941）均低分高用。
- 新血质量在改善：cohortTrend 新插件队列均分从 68.9（2026-08-17）升至 77.3（2026-10-19）。
- 头部星标与 npm 交付脱节：zhu1090093659/dsh-web 星 9512 仓库版 0.1.1 领先 npm 0.0.1。
- 高质低星是官方应抢救的存量：S 级零星插件 Shane-Jay/dsh-self-update、SenmuuuuW/dsh-whale-report 未被收录。
2. 头部插件深潜
- dsh-market/dsh-market：官方市场插件，A 级 90，周下载 175241；是生态入口，但仓库/npm 版本错位需警惕。
- Ebony-Vinyl/dsh-our-free-model：星 42984、A 级 93、周下载 7188；星标暴涨（11 天 +34108）但装机转化低，观望情绪浓。
- ranxianglei/billion-context：B 级 86、周下载 595244；高下载低星，未被 awesome/imsai 收录，是隐形工作马。
- sunyuhuirong/deepseek-account：星仅 1、B 级 76、周下载 266575；无描述、无中文文档、超 30 天无提交，高采用低维护。
- MeteorNOX/DeepSeek-Balance-Whale-Widget：星 6067、A 级 90、周下载 53699；爆款挂件催生多个低分二创（如 desktop C 级 71）。
头部格局是「入口市场 + 巨型工具箱 + 萌系挂件」三分天下，高星与高下载不重合，增长来自渠道而非质量。
3. 作者格局
- 集中度：top10Share 2.8%，星标分散，无垄断性头部（除 Ebony-Vinyl）。
- 头部作者：Ebony-Vinyl 4 插件星 43400，同时是连接者中心；Minglink 5 插件星 5789 均分仅 72.8，量质不齐。
- 多产作者：xby-skill 175 插件星仅 2、均分 62.2；PerryLink 98 插件均分 82.6——多产不等于高质。
- 高质作者：Fish-under-sea 4 插件均分 96.5 星仅 16；he0119 3 插件均分 95 星仅 2，高质作者缺曝光。
- 批量号：uckkk 1286 插件（占 8%）均分 56，CAI-MH 9 插件均分 57，污染作者榜。
当前是流量驱动 + 社区萌芽混合：头部话题引爆星标，但高质作者无渠道，批量号稀释真实贡献。
4. 质量 × 采用度交叉发现
- 被低估的宝藏：SenmuuuuW/dsh-whale-report（S,100,周152）从会话日志生成报告；Shane-Jay/dsh-self-update（S,100,周24）git 源自更新；EasyTZ/dsh-git（S,100,周55）可视 Git 面板——均未被官方收录。
- 低质高用：Creakono/dsh-cost-meter（C,64,周24941）无 LICENSE、无测试、停滞；levi52/dsh-pet（C,67,周46041）无测试/CI，滞停却高下载；LuniteGlaze/dsh-plugin-catalog（C,64,周17374）无描述。
高下载集中在即时满足场景，用户不审查维护状态；官方应默认展示风险标签。
5. 生态主题地图
- 主题一：会话与记忆（session 95、memory 68、skills 61）代表：dsh-market、dsh-skills-manager。
- 主题二：模型路由与代理（agent 57、mcp 56，模型/代理 1781）代表：dsh-routing-suite、dsh-our-free-model。
- 主题三：UI/侧栏改造（ui 73、sidebar 49、webui 51）代表：DSH-better-sidebar、dsh-wide-conversation。
- 主题四：视觉/图像与工具（vision 55、image 54、web 60）代表：liustack/modlens、dsh-imggenerate（C）。
- 拥挤方向：其它 5838 个、均分 70.5，开发/数据 494 个、均分 74.0，数量高但质量不明。
- 空白机会：记忆/知识 463 个均分 75.9、消息/协作集成 554 个均分 75.7 规模小均分高；terminal-ssh、remote-mobile 场景各仅 6 个，蓝海。
6. 生态健康度评估
| 维度 | 本期数值 | 解读 |
|---|---|---|
| 规模 | 17011（含 uckkk 1286） | 规模膨胀，需剔除批量号后看有效供给 |
| 增长 | 相对 2026-10-15 基线累计 +1494（11 天） | 最近完整周 2026-10-12 为 1422，增速仍高但水分未知 |
| 质量分布 | A 625, B 7955, C 6012, D 2284, S 135；avg 73.8 | B/C 占 82%，S 仅 0.8%，整体偏中低 |
| 活跃度 | 7 日 2759（16.2%），30 日 7673（45.1%） | 近半数仓库月内无活动，维护断层 |
| npm 发布率 | 43.7% published | 过半插件只在仓库中，安装侧可选面窄 |
| 文档覆盖 | readme 16966/17011，zh 48.3%，无文档 45 | 文档基础好，但中文覆盖不足一半 |
cohortTrend：新插件队列周均分从 68.9 升至 77.3，近 10 周持续改善，说明新供给在变优质而非变水。
7. 异常与应对
owner-cluster (mid) — uckkk 1286 插件均分 56
- 可能原因（假设）：模板复制或脚本批量生成，同一代码变形上传。
- 影响面：作者榜、规模数字被严重污染，搜索充满同质低分件。
- 应对建议：DSH 官方对 uckkk 冻结/限流，统计口径单独标记批量号。
owner-cluster (mid) — CAI-MH 9 插件均分 57
- 可能原因（假设）：小规模模板账号或半自动生成，未投入维护。
- 影响面：少量但可能占据入口，拉低品类平均分。
- 应对建议：对 avg<60 多插件账号人工审核或 shadow ban。
dominant-owner (low) — uckkk 独占 8% 均分 56
- 可能原因（假设）：与上同源，批量号规模化形成规模幻觉。
- 影响面：对外宣传如直接引用 17011 会误导投资者/用户。
- 应对建议：官方展示有效插件数，建议口径 15725（17011-1286）。
npm-stale (low) — 10 个插件仓库版本领先 npm
- 可能原因（假设）：作者忘记 publish、CI 缺失或 npm 账号所有权问题。
- 影响面：安装侧拿旧版，可能含旧漏洞，损害信任。
- 应对建议：发布页展示 repo vs npm 版本并警告，自动通知作者。
8. 下一阶段重点与建议
- 插件作者：上架前核对 npm 版本；补齐 docs/CI/test 可拉高评分；避开「其它」选高分赛道；对高下载插件设维护提醒。
- 生态用户：安装前看 score 和 dormant/no-license；高下载不等于可靠（dsh-pet、schedule 慎用）；关注 gems 中的 S 级插件；遇 npm 漂移去 GitHub release 取新版。
- DSH 官方：清理 uckkk/CAI-MH 并发布有效口径；设 60 分市场门槛+风险标签；自动推荐 gems；监控 npm-stale 并提供一键 publish。
9. 官方动态解读
- DSH 主仓库 星 261476，最新 v0.2.1-rc.3 非 breaking——迁移负担低。
- 官方模型矩阵 DeepSeek-V3/R1 星均 9 万+，持续引流插件生态。
- awesome-deepseek-integration 星 39534，但 gems 多未收录，官方精选与社区未打通。
- RC 发布节奏：候选版提示插件作者保持 API 兼容测试。
DSH Ecosystem Insights · 2026-W43（2026/10/19～2026/10/25）
Generated by the DSH Insights pipeline + AI · snapshot 2026-10-26 · heuristic evaluation, not a security audit
1. Core Conclusions
- Ecosystem scale is polluted by template farms: account uckkk alone holds 1286 plugins with avg score 56, 8% of the total, so effective supply is far below 17011.
- Adoption diverges from quality: in the top downloads list, dsh-pet (C, weekly 46041), schedule (C, weekly 29926), and dsh-cost-meter (C, weekly 24941) are low-score but heavily used.
- New blood is getting better: cohortTrend average score for new plugin cohorts rose from 68.9 (2026-08-17) to 77.3 (2026-10-19).
- Top stars and npm delivery are out of sync: zhu1090093659/dsh-web has 9512 stars but repo version 0.1.1 is ahead of npm 0.0.1.
- High-quality low-star plugins are a rescue target: S-tier zero-star plugins like Shane-Jay/dsh-self-update and SenmuuuuW/dsh-whale-report are not curated anywhere.
2. Head Plugin Deep Dive
- dsh-market/dsh-market: the official market plugin, grade A 90, weekly 175241; it is the ecosystem gateway, but repo/npm version mismatch needs watching.
- Ebony-Vinyl/dsh-our-free-model: stars 42984, grade A 93, weekly 7188; star explosion (+34108 in 11 days) but low install conversion suggests wait-and-see.
- ranxianglei/billion-context: grade B 86, weekly 595244; high downloads, low stars, not in awesome/imsai — an invisible workhorse.
- sunyuhuirong/deepseek-account: only 1 star, grade B 76, weekly 266575; no description, no Chinese docs, dormant >30 days — high adoption, low maintenance.
- MeteorNOX/DeepSeek-Balance-Whale-Widget: stars 6067, grade A 90, weekly 53699; viral widget spawned multiple low-quality derivatives (e.g., a desktop C 71).
The head landscape is divided into marketplace entry, mega toolbox, and cute widget; high stars and high downloads do not overlap, meaning growth comes from channels rather than quality.
3. Author Landscape
- Concentration: top10Share only 2.8% — stars are dispersed, no monopolistic head except Ebony-Vinyl.
- Head authors: Ebony-Vinyl holds 4 plugins with 43400 stars and is a connector hub; Minglink has 5 plugins, 5789 stars but avg only 72.8, quantity over quality.
- Prolific authors: xby-skill has 175 plugins, only 2 stars, avg 62.2; PerryLink has 98 plugins, avg 82.6 — prolific does not mean quality.
- Quality authors: Fish-under-sea has 4 plugins avg 96.5 but only 16 stars; he0119 has 3 plugins avg 95 and 2 stars — high-quality authors lack exposure.
- Template pollution: uckkk with 1286 plugins (8%) avg 56, CAI-MH with 9 plugins avg 57, distort author rankings.
The current state is a traffic-driven + emerging-community mix: viral topics drive stars, but quality authors have no channel, and template accounts dilute real contribution.
4. Quality × Adoption Cross Findings
- Underrated gems: SenmuuuuW/dsh-whale-report (S,100,weekly152) generates reports from session logs; Shane-Jay/dsh-self-update (S,100,weekly24) self-updates git installs; EasyTZ/dsh-git (S,100,weekly55) visual Git panel — none are officially curated.
- Low-quality high-use: Creakono/dsh-cost-meter (C,64,weekly24941) no LICENSE, no tests, dormant; levi52/dsh-pet (C,67,weekly46041) no tests/CI, dormant yet high downloads; LuniteGlaze/dsh-plugin-catalog (C,64,weekly17374) no description.
High downloads concentrate in instant-gratification scenarios where users ignore maintenance status; officials should show risk labels by default.
5. Ecosystem Theme Map
- Theme 1: Sessions and memory (session 95, memory 68, skills 61) e.g., dsh-market, dsh-skills-manager.
- Theme 2: Model routing and agents (agent 57, mcp 56, model/agent 1781) e.g., dsh-routing-suite, dsh-our-free-model.
- Theme 3: UI/sidebar revamp (ui 73, sidebar 49, webui 51) e.g., DSH-better-sidebar, dsh-wide-conversation.
- Theme 4: Vision/image and tools (vision 55, image 54, web 60) e.g., liustack/modlens, dsh-imggenerate (C).
- Crowded: other 5838 plugins avg 70.5, dev/data 494 plugins avg 74.0 — large but unclear quality.
- White space: memory/knowledge 463 plugins avg 75.9, messaging/collab 554 plugins avg 75.7 small but high avg; terminal-ssh, remote-mobile scenarios each have only 6 plugins — blue ocean.
6. Ecosystem Health Assessment
| Dimension | This Week | Interpretation |
|---|---|---|
| Scale | 17011 (incl. uckkk 1286) | Inflated; need to exclude template accounts for effective supply |
| Growth | +1494 cumulative vs 2026-10-15 base (11 days) | Latest full week 2026-10-12 was 1422; growth still high but water content unknown |
| Quality Distribution | A 625, B 7955, C 6012, D 2284, S 135; avg 73.8 | B/C account for 82%, S only 0.8%, overall mid-low |
| Activity | 7-day 2759 (16.2%), 30-day 7673 (45.1%) | Nearly half of repos inactive within a month — maintenance gap |
| npm Publish Rate | 43.7% published | Over half of plugins only exist in repos, narrowing install-side options |
| Docs Coverage | readme 16966/17011, zh 48.3%, no docs 45 | Docs baseline good, but Chinese coverage less than half |
cohortTrend: new plugin cohort avg score rose from 68.9 to 77.3, continuously improving over the last 10 weeks, indicating new supply is becoming better, not more diluted.
7. Anomalies and Response
owner-cluster (mid) — uckkk 1286 plugins avg 56
- Possible cause (hypothesis): template copying or script batch generation, uploaded with code variants.
- Impact: author rankings and scale figures severely polluted; search flooded with homogeneous low-score items.
- Recommendation: DSH official should freeze/rate-limit uckkk and tag template accounts in statistics.
owner-cluster (mid) — CAI-MH 9 plugins avg 57
- Possible cause (hypothesis): small-scale template account or semi-automated generation with no maintenance.
- Impact: few but may occupy entry points and drag down category average.
- Recommendation: manually review or shadow ban multi-plugin accounts with avg <60.
dominant-owner (low) — uckkk holds 8% avg 56
- Possible cause (hypothesis): same source as above, scaled template farm creates scale illusion.
- Impact: external reports quoting 17011 directly would mislead investors/users.
- Recommendation: officials should display effective plugin count; suggested figure 15725 (17011-1286).
npm-stale (low) — 10 plugins repo version ahead of npm
- Possible cause (hypothesis): authors forgot to publish, missing CI, or npm ownership issues.
- Impact: installs get old builds possibly with old vulnerabilities, damaging trust.
- Recommendation: show repo vs npm version warning on publish page and auto-notify authors.
8. Next Phase Priorities and Suggestions
- Plugin authors: check npm version before publishing; adding docs/CI/test raises scores; avoid the 'other' category and choose high-score tracks; set maintenance reminders for high-download plugins.
- Ecosystem users: before installing, check score and dormant/no-license; high downloads do not mean reliable (be cautious with dsh-pet, schedule); pay attention to S-tier plugins in gems; if npm drift occurs, fetch new version from GitHub releases.
- DSH officials: clean up uckkk/CAI-MH and publish effective count; set 60-point market gate + risk labels; auto-recommend gems; monitor npm-stale and provide one-click publish.
9. Official Dynamics Interpretation
- DSH main repo has 261476 stars, latest v0.2.1-rc.3 non-breaking — low migration burden.
- Official model matrix DeepSeek-V3/R1 each >90k stars, continuously feeding the plugin ecosystem.
- awesome-deepseek-integration has 39534 stars, but most gems are not listed, meaning official curation and community are disconnected.
- RC release cadence: candidate releases remind plugin authors to maintain API compatibility testing.