生态洞察 · 2026-W40
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
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 3 个版本，其中 2 个含 breaking 变更（dsh-v0.2.1-alpha.1, dsh-v0.2.0-rc.2）——插件作者需评估适配
DSH 生态洞察 · 2026-W40（2026/09/28～2026/10/04）
由 DSH Insights 管线 + AI 生成 · 数据快照 2026-10-05 · 启发式评估，非安全审计
核心结论
- 新增量水分显著：本周按周新增 1032 个插件，但批量账号 uckkk 以 1286 个插件、均分 56 在规模上制造虚假繁荣，实际优质供给远低于表面数字。
- 下载头部与质量错位：周下载 top1 的 billion-context 仅 532 星、B 级，而多个 S 级插件周下载不足 400，采用度由渠道而非质量主导。
- 质量分布呈金字塔但底部过厚：A/S 级合计仅 597 个（4.3%），C/D 级超 6550 个（47.6%），生态整体均分 73.8，中位星数为 0，说明大多数插件无人问津。
- 官方 breaking 变更成最大风险：dsh-v0.2.1-alpha.1 与 v0.2.0-rc.2 两个 breaking 版本一周内发布，插件作者适配窗口极短，已有 downgrade（yueyexiayu/dsh-shengcheng 分数从 84 掉落至 64）。
- npm 发布链路滞后：45.4% 发布率、10 个高星插件仓库版本领先 npm，安装侧普遍拿到旧版，影响用户体验与安全更新。
头部插件深潜
- zhu1090093659/dsh-web：Web 聚合生态入口，8367 星、B 级、周下载 248，但仓库版本 0.1.1 落后 npm 0.0.1，发布链路反向错位，高星未转化为健康迭代。
- yjh051108/dsh-routing-suite：7000 星、B 级、未发布 npm、周下载缺失——路由套件具备高认知度却无安装通道，属于“高星未落地”典型。
- dsh-market/dsh-market：官方市场，5513 星、A 级、周下载 161693，是生态内搜索与发现的核心基础设施，但自身也存在 exports["./client"] 缺失与版本漂移。
- ranxianglei/billion-context：周下载 330230 的绝对冠军，但仅 532 星、B 级；其压缩上下文解决刚需却伴随 npm 版本漂移与 client 导出缺失，用户量大但工程完整度不足。
- omdsh-dev/DSH-better-sidebar：0 星、S 级、周下载 59739，高质低星的冷启动案例，说明 S 级评分可被渠道快速放大，但自然增长乏力。
作者格局
- 集中度极低但头部失真：top10Share 仅 2.8%，看似长尾健康，但最大账号 uckkk 独占 1286 个插件（9%）、均分 56，对“作者总数 7911、多插件作者 1976”的统计形成污染。
- 头部作者如 zhu1090093659（单插件 8367 星）、yjh051108（3 插件 7170 星）：靠单个爆款或少量精品占据 star 榜，但整体贡献者网络稀疏。
- 多产作者质量两极：xby-skill 175 插件均分 62.2，omdsh-dev 67 插件均分 76.6 且总星 2008；PerryLink 42 插件均分 85 但 0 星——多产不等于高影响。
- 高质量作者集中在低星区间：EasyTZ（4 插件均分 95.5、1 星）、ice5kysl（4 插件均分 94、1 星）等证明高分与曝光脱钩。
- 判断：生态目前是“流量驱动”而非“社区驱动”——下载与星数高度集中于少数渠道占位插件，贡献者协作图关键节点（cursoragent、finiking 等）的星数来自少数连接插件，未能形成广泛互助网络。
质量 × 采用度交叉发现
- 被低估的宝藏（gems）：
- SenmuuuuW/dsh-whale-report：0 星、S 级、score 100、周下载 391，从会话事件生成日报/周报，是监控场景下极实用的工具，但无星无市场收录。
- nevertoday/dsh-theme-plugin：25 星、S 级、score 98、周下载 325，中国传统色主题包，小而美，但主题类竞争激烈且曝光不足。
- hxy91819/dsh-auth：6 星、S 级、score 98、周下载 326，Argon 认证插件，安全类需求明确却无人问津。
- 低质高用（risks）：
- Bald0Wang/dsh-imggenerate：3 星、C 级、score 62、周下载 7204，图像生成功能需求旺盛但工程完整度低，用户面临兼容与维护风险。
- LuniteGlaze/dsh-plugin-catalog：1 星、C 级、score 64、周下载 3866，插件目录类工具却自身质量堪忧，可能误导用户选择。
- johndfowler/dsh-voice：0 星、D 级、score 59、周下载 996，低质语音插件在高下载区间，说明部分用户对质量无感知。
生态主题地图
- 主流聚类：会话管理（1226 插件、均分 76.6）、模型/代理（1370、均分 76.2）、文件浏览/预览（1207、均分 76.2）构成主干；能力标签 topTags 中 session、search、ui、memory 密度最高。
- 拥挤方向：其它类 4886 插件、均分 69.8，大量未分类低质插件堆积；工具/效率 677 插件、均分 75.0，与侧栏/工作区 897、均分 76.3 相比竞争激烈。
- 空白机会：消息/协作集成 454 插件、均分 75.8，记忆/知识 375 插件、均分 75.8，开发/数据 402 插件、均分 73.9——数量远低于头部类别，且 connectors 协作节点尚未形成规模化工具链。
- 官方平台动态：dsh 官方仓库 star 达 243404，DeepSeek-V3 104515 星、DeepSeek-R1 91935 星，说明平台势能巨大，但插件生态的 llmCoverage 仅 1823/13748，能力标注覆盖率不足，主题挖掘受限。
生态健康度评估
| 维度 | 本期数值 | 解读 |
|---|---|---|
| 规模 | 13748 个插件 | 总量可观，但需剔除 uckkk 批量号后评估真实供给 |
| 增长 | 本周新增 1032（byWeek）；相对 2026-09-27 基线累计新增 1281 | 增长依然高速，但新增中低质比例高 |
| 质量分布 | A：498，B：6601，C：4534，D：2016，S：99；均分 73.8 | 底部过厚，C/D 合计 6550 个，中位星数 0 表明大部分无人使用 |
| 活跃度 | 7 日活跃 3179（23.1%），30 日活跃 6541（47.6%） | 近半数插件 30 日内有活动，但 7 日活跃不足四分之一，长尾活跃度低 |
| npm 发布率 | 45.4% 已发布，stale 3815 | 超过一半插件未走 npm 发布，且已发布中 stale 数量大 |
| 文档覆盖 | readme 13712，zh 文档 6520，zhPct 47.4% | 文档覆盖尚可，但中文文档覆盖率不足一半，与中文用户占比不匹配 |
- cohortTrend：最近 4 周新插件队列平均分从 2026-09-07 的 77.4 缓慢升至 2026-09-28 的 78.1，质量有轻微改善迹象，但仍低于 80 分健康线，且本周新增量放大至 1032，改善能否持续需观察。
异常与应对
owner-cluster（severity: mid）
- 可能原因（假设）：账号 uckkk 通过模板复制或自动化脚本批量创建 1286 个插件，以数量冲榜或测试平台边界。
- 影响面：污染作者榜与规模统计，误导用户筛选与官方决策；均分 56 拉低整体质量感知。
- 应对建议：对 uckkk 名下插件进行人工抽检，若确认模板批量，标记为低可信并限制其在市场曝光；调整统计口径，排除批量号后重新计算健康度指标。
dominant-owner（severity: low）
- 可能原因（假设）：批量号未得到及时治理，导致单一账号占比 9% 的畸形结构。
- 影响面：作者集中度指标失真，top10Share 从 2.8% 被拉低、且多插件作者统计被稀释，影响生态多样性评估。
- 应对建议：在作者维度统计中排除 uckkk 或单独标注“批量账号”，并在官方面板中增加“剔除异常账号”的开关。
npm-stale（severity: low）
- 可能原因（假设）：开发者迭代仓库后未同步发布 npm，或发布流程未自动化；也可能存在 npm 包被抢注导致版本无法更新。
- 影响面：安装侧用户获取旧版代码，可能缺少安全修复与新功能；10 个受影响插件中包括 dsh-web、dsh-market 等高星项目，影响面广。
- 应对建议：在插件提交 CI 中增加“仓库与 npm 版本一致性”检查；对高星插件定向提醒作者发布新版，并公示 stale 列表。
breaking-release（severity: high）
- 可能原因（假设）：官方在迭代 alpha/rc 阶段集中引入破坏性 API 变更，未提供足够的迁移窗口或兼容层。
- 影响面：插件作者面临紧急适配压力，已有 yueyexiayu/dsh-shengcheng 评分从 84 降至 64，可能引发连锁失效；用户依赖的插件可能在升级后不可用。
- 应对建议：官方发布 breaking 变更前提前至少 2 周公告并提供迁移指南；插件作者应订阅官方 release 并建立自动化兼容性测试；生态用户暂缓升级至 breaking 版本，等待插件适配完成。
下一阶段重点与建议
- 插件作者：
- 立即检查自身插件与 dsh-v0.2.1-alpha.1 / v0.2.0-rc.2 的兼容性，优先修复 exports["./client"] 缺失与 npm 版本漂移。
- 利用 quadrant.gems 名单中的高质低星案例（如 dsh-whale-report）分析其功能定位，寻找差异化机会。
- 避免踏入“其它”类红海，转向消息/协作、记忆/知识等低竞争类别，并主动提升中文文档覆盖。
- 生态用户：
- 安装高下载低分插件（如 dsh-imggenerate、dsh-plugin-catalog）前先检查其仓库状态与维护活跃度，优先选择 A/S 级且 30 日内有提交的插件。
- 关注 signals 中 breaking 版本公告，延迟更新等待社区反馈；对 uckkk 名下插件保持警惕。
- 利用 dsh-market 和 awesome 列表交叉验证插件可信度，不要仅凭 star 数选择。
- DSH 官方：
- 立即治理 uckkk 批量账号，发布治理结果并修订质量分算法以抑制模板复制。
- 为 breaking release 建立迁移指南与插件适配状态看板，降低生态震荡。
- 推动 npm 发布闭环，对 stale 插件自动提醒或下架处理；提升 llmCoverage 覆盖率，让主题挖掘更精准。
官方动态解读
- dsh 单周 3 个 release，其中 2 个 breaking：加速迭代意图明显，但破坏性变更集中发布对生态稳定性造成直接冲击，需评估发布节奏。
- 官方平台仓库 star 达 243404，DeepSeek-V3 与 R1 均超 9 万星：模型层势能强劲，但插件生态仅 1.3 万插件且质量分化，官方需投入更多资源做质量治理与开发者激励。
- dsd-market 周下载 161693，成为第二大下载插件：官方市场正在成为事实入口，但其自身版本漂移问题需优先修复，否则可能成为传播旧版的渠道。
- coverage 显示 topicUniverse 17603、candidates 18321、authoritative 13748：候选池大于权威池，说明存在大量未经认证或低质插件，生态准入门槛与审核机制亟待加强。
DSH Ecosystem Insights · 2026-W40（2026/09/28～2026/10/04）
Generated by the DSH Insights pipeline + AI · snapshot 2026-10-05 · heuristic evaluation, not a security audit
Core Conclusions
- Scale is inflated by batch accounts: Weekly new plugins hit 1032, but account uckkk with 1286 plugins at avg score 56 distorts supply; true quality supply is far lower than surface numbers.
- Download and quality misaligned: Top downloaded billion-context has only 532 stars and grade B, while multiple S-grade plugins get fewer than 400 weekly downloads — adoption is channel-driven, not quality-driven.
- Quality pyramid has a heavy base: A/S total only 597 (4.3%), C/D exceed 6550 (47.6%); avg score 73.8, median stars 0 show most plugins are unused.
- Official breaking changes are the biggest risk: dsh-v0.2.1-alpha.1 and v0.2.0-rc.2 released in one week, with a downgrade already observed (yueyexiayu/dsh-shengcheng from 84 to 64).
- npm pipeline lags: 45.4% publish rate, 10 top-star plugins have repo ahead of npm, users get outdated builds and miss security updates.
Head Plugin Deep Dive
- zhu1090093659/dsh-web: Web aggregation ecosystem entry, 8367 stars, grade B, weekly 248; repo 0.1.1 ahead of npm 0.0.1 — reversed versioning, high stars not converted to healthy iteration.
- yjh051108/dsh-routing-suite: 7000 stars, grade B, unpublished, no weekly data; high recognition but no install channel, typical “high-star unshipped”.
- dsh-market/dsh-market: Official market, 5513 stars, grade A, weekly 161693; core discovery infrastructure but has missing client export and version drift.
- ranxianglei/billion-context: Weekly download 330230 absolute champion, but only 532 stars, grade B; context compression solves a real pain but engineering completeness lags with npm drift and missing client export.
- omdsh-dev/DSH-better-sidebar: 0 stars, S grade, weekly 59739; high-quality cold start shows S score can be amplified by channel but organic growth is weak.
Author Landscape
- Low concentration but distorted head: top10Share only 2.8%, but uckkk alone holds 1286 plugins (9%) at avg 56, polluting author totals (7911 total, 1976 multi-plugin).
- Head authors like zhu1090093659 (1 plugin 8367 stars) and yjh051108 (3 plugins 7170 stars): rely on single hits or a few quality items to top star charts, but contribution network is sparse.
- Prolific authors split: xby-skill 175 plugins avg 62.2, omdsh-dev 67 plugins avg 76.6 with 2008 stars; PerryLink 42 plugins avg 85 but 0 stars — prolific doesn't equal impact.
- High-quality authors in low-star zone: EasyTZ (4 plugins avg 95.5, 1 star), ice5kysl (4 plugins avg 94, 1 star) prove high score and exposure are decoupled.
- Verdict: Ecosystem is traffic-driven rather than community-driven — downloads and stars concentrate on channel-occupying plugins; connector graph key nodes (cursoragent, finiking) derive stars from a few linking plugins, not a broad contributor network.
Quality × Adoption Cross Findings
- Underrated gems:
- SenmuuuuW/dsh-whale-report: 0 stars, S, score 100, weekly 391; generates daily/weekly reports from session events, highly practical for monitoring but no stars or market listing.
- nevertoday/dsh-theme-plugin: 25 stars, S, score 98, weekly 325; Chinese traditional colors theme, small but beautiful, but theme category is crowded and exposure low.
- hxy91819/dsh-auth: 6 stars, S, score 98, weekly 326; Argon authentication plugin, clear security need but ignored.
- Low-quality high-use risks:
- Bald0Wang/dsh-imggenerate: 3 stars, C, score 62, weekly 7204; image generation demand high but engineering weak, users face compatibility and maintenance risk.
- LuniteGlaze/dsh-plugin-catalog: 1 star, C, score 64, weekly 3866; catalog tool with poor own quality may mislead users.
- johndfowler/dsh-voice: 0 stars, D, score 59, weekly 996; low-quality voice plugin in high download range shows some users are quality-blind.
Ecosystem Theme Map
- Main clusters: Session management (1226 plugins, avg 76.6), model/proxy (1370, avg 76.2), file browsing/preview (1207, avg 76.2) form the backbone; top capability tags session, search, ui, memory are densest.
- Crowded directions: “其它” category 4886 plugins avg 69.8 with massive unclassified low-quality pile; tools/efficiency 677 avg 75.0 competes with sidebar/workspace 897 avg 76.3.
- White space opportunities: Messaging/collaboration 454 avg 75.8, memory/knowledge 375 avg 75.8, dev/data 402 avg 73.9 — far fewer than head categories, and connector graph has not scaled toolchains.
- Platform dynamics: dsh official repo 243404 stars, DeepSeek-V3 104515 and R1 91935 show huge platform momentum, but plugin ecosystem llmCoverage only 1823/13748, limiting theme mining.
Ecosystem Health Assessment
| Dimension | Current Value | Interpretation |
|---|---|---|
| Scale | 13748 plugins | Large, but must exclude uckkk batch for true supply |
| Growth | 1032 new this week (byWeek); +1281 cumulative vs 2026-09-27 baseline | High speed but high low-quality share |
| Quality distribution | A:498, B:6601, C:4534, D:2016, S:99; avg 73.8 | Heavy bottom, C/D 6550; median stars 0 means most unused |
| Activity | 7-day active 3179 (23.1%), 30-day active 6541 (47.6%) | Nearly half active in 30 days but 7-day below quarter, long-tail inactive |
| npm publish rate | 45.4% published, stale 3815 | Over half not on npm, and many published are stale |
| Docs coverage | readme 13712, zh docs 6520, zhPct 47.4 | Docs decent but Chinese coverage under half, mismatched with Chinese user base |
- cohortTrend: Last 4 weeks new plugin avg score rose from 77.4 (2026-09-07) to 78.1 (2026-09-28), slight quality improvement but still below 80 healthy line; with volume up to 1032, sustainability uncertain.
Anomalies and Responses
owner-cluster (severity: mid)
- Possible cause (hypothesis): Account uckkk batch-created 1286 plugins via template copy or automation to inflate counts or test platform limits.
- Impact: Pollutes author leaderboard and scale stats, misleads user filtering and official decisions; avg 56 drags overall quality perception.
- Suggestions: Manually sample uckkk plugins, mark as low-trust and limit exposure if template batch confirmed; adjust metrics to exclude batch accounts.
dominant-owner (severity: low)
- Possible cause (hypothesis): Batch account not governed timely, leading to single account 9% abnormal structure.
- Impact: Author concentration metrics distorted, top10Share lowered from 2.8%, multi-plugin author stats diluted, affecting diversity assessment.
- Suggestions: Exclude uckkk or label as “batch account” in author stats; add a toggle in official dashboard to remove anomalies.
npm-stale (severity: low)
- Possible cause (hypothesis): Devs update repo but not npm, or publish process not automated; possible npm squatting preventing version update.
- Impact: Users get old builds, missing security fixes and features; 10 affected include high-star dsh-web and dsh-market, broad impact.
- Suggestions: Add CI check for repo-npm version consistency; remind high-star plugin authors to publish and publicize stale list.
breaking-release (severity: high)
- Possible cause (hypothesis): Official concentrated breaking API changes in alpha/rc without enough migration window or compat layer.
- Impact: Plugin authors face urgent adaptation; yueyexiayu/dsh-shengcheng score dropped 84 to 64, may cascade; user plugins may break after upgrade.
- Suggestions: Announce breaking changes at least 2 weeks ahead with migration guide; plugin authors subscribe to releases and set up compat tests; users delay upgrading to breaking versions until adaption done.
Next Phase Priorities and Recommendations
- Plugin authors:
- Immediately check compatibility with dsh-v0.2.1-alpha.1 / v0.2.0-rc.2; fix missing exports["./client"] and npm version drift.
- Analyze gems list (e.g., dsh-whale-report) for functional gaps and differentiation opportunities.
- Avoid the “其它” red ocean; target messaging/collaboration, memory/knowledge; improve Chinese docs.
- Ecosystem users:
- Before installing high-download low-score plugins (dsh-imggenerate, dsh-plugin-catalog), check repo status and maintenance; prefer A/S grade with 30-day commits.
- Watch breaking release announcements; delay upgrades; be cautious with uckkk plugins.
- Cross-verify plugin trust via dsh-market and awesome lists, not stars alone.
- DSH official:
- Immediately govern uckkk batch account, publish results, revise scoring to suppress template farming.
- Build migration guide and plugin compatibility dashboard for breaking releases to reduce ecosystem shock.
- Close npm publishing loop with auto-reminders or takedowns for stale; increase llmCoverage for better theme mining.
Official Dynamics Interpretation
- dsh 3 releases in one week, 2 breaking: Fast iteration intent but concentrated breaking changes directly hit ecosystem stability; need to evaluate release cadence.
- Official platform repos 243404 stars, DeepSeek-V3 and R1 above 90k: Model layer momentum strong but plugin ecosystem only 13k plugins with quality split; need more investment in governance and incentives.
- dsh-market weekly 161693, second top download: Official market becoming de facto entry, but its own version drift must be fixed first to avoid spreading outdated builds.
- coverage: topicUniverse 17603, candidates 18321, authoritative 13748: Candidate pool larger than authoritative indicates many uncertified/low-quality plugins; entry threshold and review mechanism need strengthening.