生态洞察 · 2026-W41
重点关注 · 5 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 9%）、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 1 个版本，其中 1 个含 breaking 变更（dsh-v0.2.1-alpha.2）——插件作者需评估适配
DSH 生态洞察 · 2026-W41（2026/10/05～2026/10/11）
由 DSH Insights 管线 + AI 生成 · 数据快照 2026-10-12 · 启发式评估，非安全审计
核心结论
- 规模膨胀但质量两极：本周新增 942 个插件，但 B/C/D 级合计 890 个，S 级仅 10 个；批量账号 uckkk 独占 1286 个插件且均分 56，规模数字需剔除水分。
- 采用与质量严重背离：billion-context 周下载 547,791、deepseek-account 周下载 257,231，二者一个仅 B 级 86 分、一个仅 1 star 且无描述——用户正在无意识地使用低验证插件。
- 官方 breaking release 是最大短期风险：dsh-v0.2.1-alpha.2 含 breaking 变更，未适配插件可能立即失效。
- 质量拐点隐现：最新周队列平均分 78.6 为十周最高，但 8/17 队列低至 68.9 的劣质潮仍在生态中流通，整体提升任重道远。
头部插件深潜
- dsh-market（dsh-market/dsh-market）：官方市场插件，A 级 90 分、6,169 stars、周下载 124,710；它既是发现入口，也是仅有两个 npm 版本漂移警告的高质量标杆——官方背书使其成为生态事实标准。
- billion-context：状态/监控类，B 级 86 分、仅 901 stars 但周下载 547,791；缺少 client 导出、npm 版本漂移；高下载源于功能刚需，但工程债可能在未来升级时爆发。
- Whale-Widget（MeteorNOX/DeepSeek-Balance-Whale-Widget）：A 级 90 分、4,495 stars、周下载 47,879；把账户余额做成桌面宠物，验证了“可视化刚需 + 情感化交互”的拉新能力。
- dsh-purge（YuJunZhiXue/dsh-purge）：B 级 78 分、4,584 stars、未发布 npm、周下载为空；jailbreak 类内容带来 +2,136 星增长但无法通过正式渠道安装，属于典型的“话题型插件”，官方需关注合规与分发断层。
- dsh-routing-suite：B 级 88 分、7,020 stars、未发布 npm；作为工具/效率头部插件却无安装包，说明部分高星插件停留在源码圈，生态分发基础设施仍不完整。
作者格局
- 总作者 8,599、多插件作者 2,159、top10 作者 star 占比仅 3%，生态呈明显长尾分布，无明显头部垄断。
- 头部作者：zhu1090093659（2 插件、8,651 stars、均分 88）、yjh051108（3 插件、7,188 stars、均分 85）是 star 双雄；但 dsh-market 官方号以 1 插件、6,169 stars、均分 90 代表官方参与。
- 高产作者：xby-skill（175 插件、均分 62.2）和 PerryLink（96 插件、均分 82.7）形成鲜明对比；前者疑似批量模板，后者相对均衡。
- 批量号污染：uckkk 账号 1,286 个插件（占全生态 9%）、均分 56，CAI-MH 也有 9 个均分 57；这类账号直接推高规模和基数，使“新增插件数”失真。
- 判断：生态目前是“流量驱动”而非“社区驱动”——真正的社区连接者（Grivn、Jstn-1g 等）数量有限，而批量号贡献了规模但未贡献质量；官方需将 uckkk、xby-skill 等账号的产出与真实开发者隔离展示。
质量 × 采用度交叉发现
- 被低估的宝藏（gems）：SenmuuuuW/dsh-whale-report（S 级 100 分、0 stars、周下载 370）从会话日志生成日报/周报，能力价值远超当前露脸；sunzhentao/dsh--prompt--enhance（S 级 100 分、5 stars、周下载 657）是提示词增强，但全生态知晓度极低；EasyTZ/dsh-git（S 级 100 分、1 star、周下载 463）提供侧栏 Git 面板，属高需求场景。这些应被官方收录或社区推荐。
- 低质高用（risks）：Bald0Wang/dsh-imggenerate（C 级 62 分、周下载 6,370）集成 qwen-image，但质量短板可能导致生成失败或安全风险；LuniteGlaze/dsh-plugin-catalog（C 级 64 分、周下载 3,272）作为目录类插件居然缺乏基本文档，可能误导用户；johndfowler/dsh-voice（D 级 59 分、周下载 945）语音功能存在明显工程缺陷。用户安装这些插件前应查看分级与维护状态。
生态主题地图
- 会话/记忆/代理：session 95、memory 68、agent 57、skills 61 构成最密集方向，代表插件如 dsh-session-watch、dsh-memory、dsh-skills-manager。
- 搜索/连接/MCP：search 85、mcp 56、web 60，dsh-market、dsh-mcp-setting、dsh-browser 是典型；该方向与官方市场高度相关。
- UI/侧栏/桌面：ui 73、sidebar 49、webui 51，dsh-better-sidebar、dsh-codex-ui、dsh-plugin-square 竞争激烈，但 S 级稀缺。
- 视觉/图像：vision 55、image 54，modlens、dsh-wallpaper-engine、dsh-imggenerate 等，质量参差。
- 拥挤与空白：categoryMap 中“其它”占 5,242 个且均分仅 70.1，是最拥挤且低质的池子；而“消息/协作集成”仅 492 个、“记忆/知识”412 个，与高频能力标签（memory、agent）不匹配，存在结构化机会。空白机会：终端/SSH、远程/移动场景各仅 6 个插件但需求在增多，值得官方布局。
生态健康度评估
| 维度 | 本期数值 | 解读 |
|---|---|---|
| 规模 | 14,866 权威插件 | 总量大，但权威覆盖率 = 14,866/18,760 主题宇宙，仍有约 21% 候选未纳入 |
| 增长 | 基线 2026-10-04 累计 +1,265；本周新队列 942 | 周新增回归千级，但其中 B/C/D 级占 943/942（几乎全部），S 级仅 10 个 |
| 质量分布 | A 497、B 7,125、C 5,053、D 2,094、S 97；均分 73.7 | B/C 占比近 82%，质量阈值偏低，大量“可用但不优秀” |
| 活跃度 | 7 日活跃 18%、30 日活跃 45.9% | 意味着 54.1% 插件 30 天内无任何活动，半休眠生态 |
| npm 发布率 | 44.3% 已发布；npmStaleTop 10 个版本漂移 | 超半数插件无法直接安装，且发布侧与仓库不同步 |
| 文档覆盖 | readme 14,827/14,866；zh 文档 47.8%；无文档 39 | 基本文档齐全，但中文文档不满一半，国际化或本地化仍有短板 |
cohortTrend 解读：最近十周队列平均分从 75.5（08-03）跌至 68.9（08-17）后持续回升，最新 10-05 队列达 78.6，为十周最高——说明新插件质量在改善，但 8 月中旬涌入的劣质插件已沉淀在库中，拉低整体平均分；官方应加强回溯清理，而非仅看新队列。
异常与应对
owner-cluster · mid · uckkk 批量号
- 可能原因（假设）：模板复制、自动化脚本批量生成插件，追求数量而非质量。
- 影响面：1,286 个均分 56 的插件污染榜单、搜索、分类统计，掩盖真实生态质量。
- 应对建议：DSH 官方对 uckkk 账号实施限流、复核前隐藏其插件，并在统计口径中标注“剔除批量号”。
owner-cluster · mid · CAI-MH 批量号
- 可能原因（假设）：同模板或低质复制，9 个插件均分 57、1 个无 README。
- 影响面：虽规模较小，但若未被管控可能演变为另一批量号，污染特定分类。
- 应对建议：官方核查 CAI-MH 的 9 个插件，对无 README 的要求补全，否则降权或下架。
dominant-owner · low · uckkk 规模垄断
- 可能原因（假设）：批量号利用平台注册审核宽松，快速铺量形成长尾头部。
- 影响面：uckkk 独占 9% 插件数，但均分 56，使作者榜、插件数等指标失真，用户可能被大量低质内容淹没。
- 应对建议：在公开榜单中单列“批量/疑似模板”作者，并在搜索默认结果中降低该类插件权重。
npm-stale · low · 版本漂移
- 可能原因（假设）：作者发布后未同步 npm，或 npm 包被他人抢注/错配导致 latest 落后于仓库。
- 影响面：用户通过 npm 安装到旧版本，可能缺少修复或功能，甚至与当前 DSH 不兼容。
- 应对建议：官方提供 npm 同步检测工具，对漂移插件在详情页显示“版本落后”警告，并推动作者一键重新发布。
breaking-release · high · dsh-v0.2.1-alpha.2
- 可能原因（假设）：官方为修复架构或安全缺陷而主动引入 breaking 变更，属正常迭代但风险高。
- 影响面：所有依赖旧 API 的插件可能在升级后失效，尤其是未发布 npm 或未持续维护的插件。
- 应对建议：插件作者立即对照 changelog 检查自己的插件，官方可发布迁移指南并给出一段宽限期；用户需暂缓升级到 alpha.2，等待稳定版。
下一阶段重点与建议
- 插件作者：1. 立即检查 dsh-v0.2.1-alpha.2 的 breaking 变更并更新插件；2. 修复 npm 版本漂移，确保 npm latest 与仓库同步；3. 补充 client 导出与 docs/ 目录，消除 warn 级问题；4. 主动对照 signal 与风险清单，避免因低质被下架。
- 生态用户：1. 安装前查看插件 grade 与 drops 列表，避开 D/C 级高下载插件；2. 优先选择 published 且 npm 版本一致的插件；3. 对 0 star 但 S 级 gem 给予关注，可先小范围试用；4. 切勿在生产环境直接升级 alpha.2，等稳定版。
- DSH 官方：1. 立即冻结 uckkk 与 CAI-MH 账号的新建权限，清查历史批量插件；2. 在仪表盘中分离“真实作者”和“批量号”，统计时剔除以避免失真；3. 建立 npm 同步自动检测和告警，对 stale 超过 30 天的插件置顶提醒；4. 将 quadrant.gems 中的高质低用插件纳入官方 featured 或 awesome 列表，促进质量正向循环。
官方动态解读
- dsh-v0.2.1-alpha.2 发布（breaking）：官方以 alpha 版本引入 breaking 变更，说明架构正在调整，插件作者需跟随，否则会被甩下；对生态的短期影响是适配成本上升。
- dshStars 达 247,518，官方推送日 2026-10-09：官方仓库活跃度尚可，但与插件生态巨大规模相比，核心开发速度仍显保守。
- 平台仓库 DeepSeek-V3 104,499 stars、R1 91,905 stars：上游模型热度持续反哺 DSH 生态，新插件中模型/代理类占 1,489 个，未来增长可期。
- awesome-deepseek-integration 39,348 stars：集成生态受关注，但 DSH 权威插件仅 14,866，说明大量集成尚未插件化，官方应加大对底层能力封装的支持。
DSH Ecosystem Insights · 2026-W41（2026/10/05～2026/10/11）
Generated by the DSH Insights pipeline + AI · snapshot 2026-10-12 · heuristic evaluation, not a security audit
Core Conclusions
- Scale inflates while quality splits: 942 new plugins this week, but B/C/D grades total 890 and only 10 S-grade; batch account uckkk holds 1,286 plugins with avg score 56, so scale metrics need de-noising.
- Adoption and quality diverge sharply: billion-context has 547,791 weekly downloads but only B-grade 86; deepseek-account has 257,231 downloads but 1 star and no description — users are unknowingly installing under-validated plugins.
- Official breaking release is the largest short-term risk: dsh-v0.2.1-alpha.2 contains breaking changes; unadapted plugins may fail immediately after upgrade.
- A quality turn may be emerging: latest weekly cohort avg score 78.6 is the highest in ten weeks, but the 68.9 low-quality wave from 08-17 remains in circulation; overall recovery is slow.
Top Plugin Deep Dive
- dsh-market (dsh-market/dsh-market): Official marketplace plugin, A-grade 90, 6,169 stars, 124,710 weekly downloads; it is both discovery entry and quality benchmark with only two npm drift warnings — official endorsement makes it the de facto standard.
- billion-context: Status/monitoring category, B-grade 86, only 901 stars but 547,791 weekly downloads; missing client export and npm drift; high downloads driven by basic need, but engineering debt may erupt in future upgrades.
- Whale-Widget (MeteorNOX/DeepSeek-Balance-Whale-Widget): A-grade 90, 4,495 stars, 47,879 weekly downloads; turning account balance into a desktop pet validates the pull of “visualized utility + emotional interaction”.
- dsh-purge (YuJunZhiXue/dsh-purge): B-grade 78, 4,584 stars, unpublished on npm, zero weekly downloads; jailbreak content brought +2,136 stars but cannot be installed via official channel — a typical "topic-driven plugin"; official must address compliance and distribution gap.
- dsh-routing-suite: B-grade 88, 7,020 stars, unpublished on npm; a top tool/efficiency plugin without install package indicates high-star plugins remain in source-only circles; distribution infrastructure is incomplete.
Author Landscape
- Total authors 8,599, multi-plugin authors 2,159, top 10 authors share only 3% of stars — a clear long-tail distribution with no dominant oligopoly.
- Head authors: zhu1090093659 (2 plugins, 8,651 stars, avg 88) and yjh051108 (3 plugins, 7,188 stars, avg 85) are the star duo; dsh-market official account represents official participation with 1 plugin, 6,169 stars, avg 90.
- Prolific authors: xby-skill (175 plugins, avg 62.2) and PerryLink (96 plugins, avg 82.7) contrast sharply; the former is suspected template farming, the latter relatively balanced.
- Batch account pollution: uckkk holds 1,286 plugins (9% of ecosystem) with avg 56; CAI-MH has 9 plugins avg 57; these accounts directly inflate totals and base counts, distorting “new plugin count”.
- Judgment: The ecosystem is currently “traffic-driven” rather than “community-driven” — real connectors (Grivn, Jstn-1g, etc.) are few, while batch accounts contribute scale but not quality; official isolation of batch accounts from real developers is needed.
Quality × Adoption Cross-Findings
- Undervalued gems: SenmuuuuW/dsh-whale-report (S-grade 100, 0 stars, weekly 370) generates daily/weekly reports from session logs; sunzhentao/dsh--prompt--enhance (S 100, 5 stars, 657) is prompt enhancement with extremely low visibility; EasyTZ/dsh-git (S 100, 1 star, 463) provides a sidebar Git panel for a high-demand scenario. These should be featured or recommended by the community.
- Low-quality high-usage risks: Bald0Wang/dsh-imggenerate (C 62, weekly 6,370) integrates qwen-image but may fail or create security risks; LuniteGlaze/dsh-plugin-catalog (C 64, weekly 3,272) is a catalog plugin lacking basic docs; johndfowler/dsh-voice (D 59, weekly 945) has obvious engineering flaws. Users should check grade and maintenance status before installing these.
Ecosystem Theme Map
- Session / Memory / Agent: session 95, memory 68, agent 57, skills 61 form the densest direction; representatives: dsh-session-watch, dsh-memory, dsh-skills-manager.
- Search / Connection / MCP: search 85, mcp 56, web 60; dsh-market, dsh-mcp-setting, dsh-browser are typical; strongly tied to official market.
- UI / Sidebar / Desktop: ui 73, sidebar 49, webui 51; dsh-better-sidebar, dsh-codex-ui, dsh-plugin-square compete intensely but S-grade is scarce.
- Vision / Image: vision 55, image 54; modlens, dsh-wallpaper-engine, dsh-imggenerate; quality uneven.
- Crowded and empty spaces: “Other” category has 5,242 plugins with avg score 70.1, the largest low-quality pool; “Messaging / Collaboration” only 492 and “Memory / Knowledge” only 412, mismatched with high-frequency tags (memory, agent) — structured opportunity exists. Blank opportunities: terminal/SSH and remote/mobile scenarios have only 6 plugins each but demand is rising, worth official attention.
Ecosystem Health Assessment
| Dimension | Current Value | Interpretation |
|---|---|---|
| Scale | 14,866 authoritative plugins | Large total, but authoritative coverage = 14,866/18,760 topic universe, ~21% candidates still unincorporated |
| Growth | Baseline 2026-10-04 cumulative +1,265; this week new cohort 942 | Weekly additions return to thousand-level, but B/C/D grades account for nearly all; only 10 S-grade |
| Quality Distribution | A 497, B 7,125, C 5,053, D 2,094, S 97; avg 73.7 | B/C share ~82%; quality threshold low, many “usable but not excellent” |
| Activity | 7-day active 18%, 30-day active 45.9% | 54.1% plugins have no activity in 30 days — semi-dormant ecosystem |
| npm Publish Rate | 44.3% published; npmStaleTop 10 version drift | More than half cannot be installed directly; publishing side out of sync with repo |
| Documentation Coverage | readme 14,827/14,866; zh docs 47.8%; no docs 39 | Basic docs complete, but Chinese docs less than half; localization still weak |
cohortTrend interpretation: Average score of recent ten weekly cohorts fell from 75.5 (08-03) to 68.9 (08-17), then continuously recovered to 78.6 in the 10-05 cohort, the highest in ten weeks — new plugin quality is improving, but the low-quality wave from mid-August has settled in the library, dragging down overall average; official should strengthen retrospective cleanup, not just look at new cohorts.
Anomalies and Responses
owner-cluster · mid · uckkk batch account
- Possible cause (hypothesis): Template copying, automated script batch generation, pursuing quantity over quality.
- Impact: 1,286 plugins with avg 56 pollute leaderboards, search, category stats, hiding true ecosystem quality.
- Recommendation: DSH official should rate-limit uckkk, hide its plugins pending review, and label statistics as “excluding batch accounts”.
owner-cluster · mid · CAI-MH batch account
- Possible cause (hypothesis): Same template or low-quality duplication; 9 plugins avg 57, 1 without README.
- Impact: Small scale, but if uncontrolled may become another batch account, polluting specific categories.
- Recommendation: Official review CAI-MH’s 9 plugins; require README completion or downgrade/removal.
dominant-owner · low · uckkk scale monopoly
- Possible cause (hypothesis): Batch account exploits loose registration review, rapidly scales to form long-tail head.
- Impact: uckkk alone holds 9% of plugins with avg 56, distorting author rankings and plugin counts; users may be flooded with low-quality content.
- Recommendation: Separate “batch/suspected template” authors in public lists, and lower such plugins’ weight in default search results.
npm-stale · low · version drift
- Possible cause (hypothesis): Authors did not sync npm after release, or npm package was squatted/mismatched causing latest to lag repo.
- Impact: Users install outdated versions via npm, missing fixes/features, possibly incompatible with current DSH.
- Recommendation: Official provide npm sync detection tool, show “version behind” warning on detail page, and push authors to one-click republish.
breaking-release · high · dsh-v0.2.1-alpha.2
- Possible cause (hypothesis): Official intentionally introduced breaking changes to fix architecture/security, normal iteration but high risk.
- Impact: All plugins relying on old APIs may fail after upgrade, especially those unpublished or unmaintained.
- Recommendation: Plugin authors immediately check changelog and update plugins; official publish migration guide and grace period; users should delay upgrading to alpha.2 until stable.
Next Period Priorities and Suggestions
- Plugin authors: 1. Immediately check dsh-v0.2.1-alpha.2 breaking changes and update plugins; 2. Fix npm version drift, ensure npm latest synced with repo; 3. Add client export and docs/ directory to clear warn-level issues; 4. Proactively review signals and risk lists to avoid removal.
- Ecosystem users: 1. Check plugin grade and drops before installing, avoid D/C high-download plugins; 2. Prefer published plugins with consistent npm versions; 3. Pay attention to 0-star S-grade gems, try in small scope first; 4. Do not upgrade to alpha.2 in production, wait for stable.
- DSH official: 1. Immediately freeze new-plugin permissions for uckkk and CAI-MH, audit historical batch plugins; 2. Separate “real authors” and “batch accounts” in dashboard, exclude from stats to avoid distortion; 3. Build npm sync auto-detection and alerting, highlight plugins stale over 30 days; 4. Feature quadrant.gems high-quality low-usage plugins in official featured/awesome lists to promote quality cycle.
Official Dynamics Interpretation
- dsh-v0.2.1-alpha.2 released (breaking): Official alpha with breaking changes indicates architecture adjustment; plugin authors must follow or be left behind; short-term adaptation cost rises.
- dshStars 247,518, official push 2026-10-09: Official repo activity decent, but relative to huge plugin ecosystem, core development pace still conservative.
- Platform repos DeepSeek-V3 104,499 stars, R1 91,905 stars: Upstream model heat continuously feeds DSH ecosystem; new plugins in model/agent category now 1,489, future growth promising.
- awesome-deepseek-integration 39,348 stars: Integration ecosystem attracts attention, but DSH authoritative plugins only 14,866, meaning many integrations not yet pluginized; official should strengthen support for underlying capability encapsulation.