生态洞察 · 2026-W38
重点关注 · 5 个异常信号
中本周新增 1602，近 3 周均值 219——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
低最大账号 uckkk 独占 1286 个插件（占全生态 11%）、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 2 个版本，其中 2 个含 breaking 变更（dsh-v0.1.6-alpha.2, dsh-v0.1.6-alpha.1）——插件作者需评估适配
DSH 生态洞察 · 2026-W38（2026/09/14～2026/09/20）
由 DSH Insights 管线 + AI 生成 · 数据快照 2026-09-21 · 启发式评估，非安全审计
核心结论
- 规模注水明显：新增 1602 个插件（基线 2026-09-08，13 天累计），但仅 uckkk 一个账号就贡献 1286 个、均分 56，剔除后真实增长有限。
- 赢家通吃加剧：下载量前十的插件中，6 个来自同一作者 MichengAI 或头部组织，而大量 S 级高质插件 star 与下载近乎为零。
- npm 发布链断裂：npmStaleTop 中 dsh-market 仓库版本 1.43.0 领先 npm 1.52.0，用户安装到的是过期构建。
- 官方 breaking 变更迫在眉睫：dsh-v0.1.6-alpha.2 发布含 breaking，所有插件作者需立即评估适配。
- 新队列质量回升：cohortTrend 最新周 2026-09-14 平均分 78.6，较此前低谷 69.9 明显反弹，生态有从“数量堆砌”转向“质量筛选”的迹象。
头部插件深潜
- dsh-market/dsh-market：插件市场入口，star 4262、grade A、weekly 89707，但仓库版本 1.43.0 落后 npm 最新 1.52.0，且存在 manifest.no-client-export 警告——高下载与版本漂移并存，用户实际体验可能低于评分。
- omdsh-dev/DSH-better-sidebar：开放侧边栏底座，star 0、grade S、score 95、weekly 52147；零星却高下载，说明该插件的价值由官方分发或默认集成驱动，但作者未获得社区认可，值得关注其后续维护。
- MichengAI 系列（dsh-codex-ui / dsh-agency-agents / dsh-archive-manager）：作者一人覆盖侧栏、模型代理、会话管理，stars 57–85、grade A/B、weekly 2.2 万–2.6 万，均入列 inAwesome；系列化打法清晰，是典型的高质连接者。
- bowenliang123/dsh-context：上下文管理，star 1453、grade A、weekly 22115，但 npm 落后（0.42.0 vs 0.54.0），且缺失 files 目录——高星插件同样存在发布与文档缺陷。
- liustack/modlens：视觉桥接工具，star 4000、grade B、weekly 15389，npm 版本 3.25.4 vs 3.26.2 漂移；头部作者对发布链路的疏忽普遍存在。
作者格局
- 集中度极低：authors.top10Share 仅 4%，总作者 6742、多插件作者 1624，说明 star 高度分散，没有霸主，但批量号拉高了“作者总数”的虚胖。
- 头部作者：zhu1090093659（dsh-web，7861 stars，未发布 npm）与 yjh051108（dsh-routing-suite，7199 stars，未发布）占据 star 榜前二，但均未发布 npm，属于“展示型”引流而非可安装插件。
- 多产作者：omdsh-dev 以 67 个插件、平均 77.1 分、总 1878 stars 成为高产高质代表；xby-skill 175 个插件却只有 1 star、均分 67.2，疑似批量模板。
- 高质作者：EasyTZ（4 插件，平均 95.5 分，仅 1 star）与 ice5kysl（4 插件，平均 95.3 分，0 star）是典型“高质无人问”，反衬生态的流量分配问题。
- 批量号污染：uckkk 一个账号 1286 个插件、均分 56，占生态 11%，严重干扰作者榜和规模统计；官方需将其标记并剔除。
- 判断：生态目前呈“流量驱动”而非“社区驱动”——下载和 star 集中在少数入口型插件，真正高质量的个人作者被淹没。
质量 × 采用度交叉发现
- 被低估的宝藏：
- SenmuuuuW/dsh-whale-report：报表生成，S 级 100 分、weekly 127、star 0；功能完整但完全未被发现。
- y2zyyr/dsh-token-usage-sidebar：Token 用量侧栏，S 级 100 分、weekly 135、star 3；对成本敏感用户极具价值。
- EasyTZ/dsh-git：Git 面板，S 级 100 分、weekly 160、star 1；生产力工具却无人问津。
- 低质高用：
- genius-alray/dsh-mermaid-render：C 级 64 分、weekly 17945、star 1、无描述；高下载依赖默认推送，用户可能安装到劣质实现。
- Creakono/dsh-cost-meter：C 级 64 分、weekly 17903、star 1；成本计算功能透明度低，存在误导风险。
- ben7am1n/dsh-browser：C 级 61 分、weekly 721、star 6；浏览器自动化插件质量不足，易被低分替代。
- 判断：值得官方在市场中加入“被低估宝藏”推荐位，同时对低分高下载插件标注风险。
生态主题地图
- 工作区 / 侧栏增强：代表 omdsh-dev/DSH-better-sidebar、MichengAI/dsh-codex-ui、EasyTZ/dsh-git；该方向插件数 714、平均 77.4 分，竞争激烈但优质供给充足。
- 模型路由 / 代理：代表 yjh051108/dsh-routing-suite、biyuhao/dsh-model-proxy、MichengAI/dsh-agency-agents；模型/代理类 1142 个、平均 77 分，是核心赛道但专业门槛高。
- 会话 / 记忆 / 上下文：代表 bowenliang123/dsh-context、MichengAI/dsh-archive-manager；会话管理 998 个、记忆/知识 321 个，平均分 77.1/76.9，需求稳固，但缺少杀手级产品。
- 监控 / 用量 / 余额：代表 MeteorNOX/DeepSeek-Balance-Whale-Widget、y2zyyr token-usage-sidebar；状态/监控 822 个、平均 77.6 分，鲸鱼娘等高辨识度插件出现，用户心智逐渐形成。
- 拥挤且低质：“其它”类占 4315 个、平均分仅 69.7，是批量模板重灾区；“主题/视觉美化”452 个平均 74.8，“开发/数据”353 个平均 74.5，存在大量低质竞争。
- 空白机会：LLM 能力标注覆盖率仅 1823/11792，导致主题地图不完整；MCP 集成（tag 56）、多模态 vision（55）、image（54）等方向供给不足，可作为新切入点。
生态健康度评估
| 维度 | 本期数值 | 解读 |
|---|---|---|
| 规模 | authoritative 11792 | 总量看似庞大，但含 uckkk 等批量号，去重后约 10506；真实规模需重新校准。 |
| 增长 | +1602（相对 2026-09-08，13 天）；单周 09-14 桶 639 | 基期非标准周，不能视为“本周新增”；增长主要由批量导入驱动，自然增长有限。 |
| 质量分布 | S 156、A 617、B 5569、C 3698、D 1752；avgScore 74.1 | B/C 占 78.6%，高质量 S/A 仅 6.5%；平均分被大量 C/D 拖累，整体质量中下。 |
| 活跃度（7 日 / 30 日） | active7 2256（19.1%）、active30 6813（57.8%） | 五分之四插件近 7 天无活跃，生态活跃层极薄，多数插件发布后即静止。 |
| npm 发布率 | publishPct 44.5%；stale 2698 | 超半数插件未发布 npm，安装依赖仍以 git 源为主，版本管理混乱。 |
| 文档覆盖 | readme 11763、zh 5619；zhPct 47.6%；none 29 | 中文文档覆盖率不足一半，且存在双语不一致；无文档极少数但影响发现。 |
cohortTrend 解读：新插件队列平均分经历 2026-08-17 的 69.9 低谷后，连续三周回升至 78.6，说明低质批量涌入的峰值已过，近期新作者更偏向优质实现；但样本量从 4013 降至 639，需观察是否可持续。
异常与应对
growth-anomaly（severity: mid）
- 可能原因（假设）：批量导入或刷库导致短时大量新增，并非自然开发行为。
- 影响面：规模统计失真，总插件数、作者数均被注水，掩盖真实活跃度与质量趋势。
- 应对建议：官方过滤批量账号并发布去水口径；仪表盘发布阶段新增“剔除异常后新增”指标。
owner-cluster（severity: mid）
- 可能原因（假设）：账号 uckkk 通过模板批量生成插件，缺乏实质功能与维护。
- 影响面：推高插件总数、拉低平均分，用户在搜索和榜单中可能被低质内容淹没。
- 应对建议：将 uckkk 标记为批量号并降权；触发人工审核，必要时下架其插件。
dominant-owner（severity: low）
- 可能原因（假设）：单一账号大规模注册，与 owner-cluster 同源，但侧重占比影响。
- 影响面：uckkk 占全生态 11%，任何作者榜、活跃榜都会被扭曲，破坏公平性。
- 应对建议：建立作者占比自动预警，单账号插件数超过阈值（如 100）自动进入审核队列。
npm-stale（severity: low）
- 可能原因（假设）：发布流水线未同步仓库，或作者未及时 npm publish。
- 影响面：用户安装侧拿到旧版，可能缺功能或含已知 bug；版本漂移破坏依赖解析。
- 应对建议：官方在市场中对仓库版本与 npm 不一致的插件加警告标记；插件作者配置自动发布。
breaking-release（severity: high）
- 可能原因（假设）：官方在 alpha 阶段引入 API 变更，dsh-v0.1.6-alpha.2 与 alpha.1 均含 breaking。
- 影响面：依赖旧接口的插件可能崩溃或行为异常，尤其影响 manifest.no-client-export 类插件。
- 应对建议：官方发布迁移指南和 changelog；插件作者对照检查 client export；用户在升级前备份配置。
下一阶段重点与建议
- 插件作者：
- 立即执行
npm publish同步仓库最新版本，消除 npm-staleTop 中的漂移。 - 对照 dsh-v0.1.6-alpha.2 changelog 检查 manifest 与 client 入口，修复 no-client-export 警告。
- 为低星高质插件补上中文 readme、topics 和 demo 动图，提升可发现性。
- 避免模仿批量模板，专注单点功能并持续维护。
- 立即执行
- 生态用户：
- 优先选择 grade S/A 且 weekly >1000 的插件，如 dsh-market、omdsh-dev/DSH-better-sidebar。
- 安装前核对仓库版本与 npm 版本是否一致，避免安装过期构建。
- 警惕下载量高但 grade C 的插件（如 genius-alray/dsh-mermaid-render），可等待官方风险提示。
- 关注 inAwesome/inImsai 标记，但也要主动探索 quadrant.gems 中的高质低星插件。
- DSH 官方：
- 清理 uckkk、xby-skill 等批量账号，将规模统计口径改为“去重 + 质量加权”。
- 在市场中增加“被低估宝藏”推荐位，并对低分高下载插件标注“质量存疑”。
- 扩大 LLM 能力标注覆盖率（当前 1823/11792），完善主题地图与搜索聚类。
- 针对 breaking release 发布迁移指南，并设置至少 2 周的适配窗口。
官方动态解读
- dsh 最新版本 dsh-v0.1.6-alpha.2（2026-09-17）含 breaking：插件作者需立即适配，否则用户升级后大量插件失效。
- 官方仓库 dshStars 231221、dshPushed 2026-09-17：平台本身热度持续，但插件生态未完全承接，大量用户仍未迁移到插件体系。
- 平台模型仓库热度高：DeepSeek-V3（104472 stars）、R1（91984）等核心模型表现强劲，插件生态需借助模型热度引导用户安装插件。
- coverage 中 authoritative 11792 低于 candidates 16294：仍存在 4500+ 候选插件未纳入官方库，官方需扩展收录范围以反映真实生态。
DSH Ecosystem Insights · 2026-W38（2026/09/14～2026/09/20）
Generated by the DSH Insights pipeline + AI · snapshot 2026-09-21 · heuristic evaluation, not a security audit
Core Conclusions
- Inflated scale: 1602 new plugins added relative to baseline 2026-09-08 (13-day span), but account uckkk alone contributed 1286 plugins with an average score of 56; real growth is limited after removal.
- Winner-takes-all: among top downloads, 6 come from author MichengAI or leading orgs; many S-grade high-quality plugins have near-zero stars and downloads.
- npm pipeline broken: dsh-market repo version 1.43.0 leads npm 1.52.0, so users install outdated builds.
- Official breaking change imminent: dsh-v0.1.6-alpha.2 includes breaking changes; all plugin authors must adapt immediately.
- New cohort quality recovering: latest cohort week 2026-09-14 average score 78.6, up from 69.9 trough, indicating a possible shift from quantity to quality.
Head Plugin Deep Dive
- dsh-market/dsh-market: market entrance, 4262 stars, grade A, weekly 89707; but repo v1.43.0 vs npm 1.52.0 and manifest.no-client-export warning — high downloads co-exist with version drift.
- omdsh-dev/DSH-better-sidebar: open sidebar foundation, 0 stars, grade S, score 95, weekly 52147; zero stars yet high downloads suggests official/default distribution without community recognition.
- MichengAI series (dsh-codex-ui / dsh-agency-agents / dsh-archive-manager): covers sidebar, model proxy, session management; 57–85 stars, grade A/B, weekly 22k–26k, all inAwesome; systematic high-quality connector.
- bowenliang123/dsh-context: context management, 1453 stars, grade A, weekly 22115; npm lags (0.42.0 vs 0.54.0) and missing files directory — high-star plugin also has publishing and docs flaws.
- liustack/modlens: vision bridge, 4000 stars, grade B, weekly 15389, npm drift 3.25.4 vs 3.26.2; leading authors commonly neglect release pipelines.
Author Landscape
- Extremely low concentration: authors.top10Share only 4%, 6742 total authors, 1624 multi-plugin authors; stars are highly dispersed but batch accounts inflate author counts.
- Top authors: zhu1090093659 (dsh-web, 7861 stars, unpublished) and yjh051108 (dsh-routing-suite, 7199 stars, unpublished) lead star chart but not published — showcase traffic rather than installable plugins.
- Prolific authors: omdsh-dev with 67 plugins, avg 77.1, 1878 stars is productive and high-quality; xby-skill with 175 plugins, 1 star, avg 67.2 is likely template batch.
- High-quality authors: EasyTZ (4 plugins, avg 95.5, 1 star) and ice5kysl (4 plugins, avg 95.3, 0 stars) are “high quality, nobody cares”, reflecting flow distribution problems.
- Batch pollution: uckkk holds 1286 plugins, avg 56, 11% of ecosystem, severely distorts author and scale stats; official must mark and exclude.
- Verdict: ecosystem is “traffic-driven” not “community-driven” — downloads and stars concentrate on a few entrance plugins, while true quality authors are drowned.
Quality × Adoption Cross Findings
- Underrated gems:
- SenmuuuuW/dsh-whale-report: report generator, S 100, weekly 127, 0 stars; complete but undiscovered.
- y2zyyr/dsh-token-usage-sidebar: token usage sidebar, S 100, weekly 135, 3 stars; high value for cost-sensitive users.
- EasyTZ/dsh-git: Git panel, S 100, weekly 160, 1 star; productivity tool ignored.
- Low quality high usage:
- genius-alray/dsh-mermaid-render: C 64, weekly 17945, 1 star, no description; high downloads via default push, risk of poor implementation.
- Creakono/dsh-cost-meter: C 64, weekly 17903, 1 star; cost calculation opacity risks misleading users.
- ben7am1n/dsh-browser: C 61, weekly 721, 6 stars; browser automation quality insufficient, easily replaced.
- Verdict: official should add “underrated gems” recommendation slots and risk warnings for low-score high-download plugins.
Ecosystem Theme Map
- Workspace / sidebar enhancement: reps omdsh-dev/DSH-better-sidebar, MichengAI/dsh-codex-ui, EasyTZ/dsh-git; 714 plugins, avg 77.4, competitive but quality supply adequate.
- Model routing / proxy: reps yjh051108/dsh-routing-suite, biyuhao/dsh-model-proxy, MichengAI/dsh-agency-agents; 1142 plugins, avg 77, core but high technical bar.
- Session / memory / context: reps bowenliang123/dsh-context, MichengAI/dsh-archive-manager; session 998, memory 321, avg 77.1/76.9, steady demand but no killer product.
- Monitoring / usage / balance: reps MeteorNOX/DeepSeek-Balance-Whale-Widget, y2zyyr token-usage-sidebar; 822 plugins avg 77.6, identifiable mascots emerge.
- Crowded and low-quality: “Other” category 4315 plugins avg 69.7 is batch template disaster; “Theme/Visual” 452 avg 74.8, “Dev/Data” 353 avg 74.5.
- Gap opportunities: LLM capability tagging coverage only 1823/11792, incomplete theme map; MCP integration (tag 56), multimodal vision (55), image (54) supply insufficient.
Ecosystem Health Assessment
| Dimension | Current Value | Interpretation |
|---|---|---|
| Scale | authoritative 11792 | large but includes batch accounts; ~10506 after dedup; real scale needs recalibration. |
| Growth | +1602 (vs 2026-09-08, 13 days); single week 09-14 bucket 639 | baseline non-standard week, cannot call “this week”; growth driven by batch imports. |
| Quality distribution | S 156, A 617, B 5569, C 3698, D 1752; avgScore 74.1 | B/C account for 78.6%, S/A only 6.5%; average pulled down by C/D. |
| Activity (7d/30d) | active7 2256 (19.1%), active30 6813 (57.8%) | 4/5 plugins inactive in 7d; thin active layer, most stop after release. |
| npm publish rate | publishPct 44.5%; stale 2698 | over half unpublished; installs rely on git sources, version chaos. |
| Documentation | readme 11763, zh 5619; zhPct 47.6%; none 29 | Chinese docs under half; few no-docs but hurt discoverability. |
cohortTrend interpretation: new plugin cohort average score rebounded from 69.9 on 2026-08-17 to 78.6 on 2026-09-14, indicating low-quality batch peak passed and recent authors favor quality; but sample size dropped from 4013 to 639, watch sustainability.
Anomalies & Responses
growth-anomaly (severity: mid)
- Possible cause (hypothesis): batch import/farming causing sudden spike, not natural development.
- Impact: scale stats distorted, totals and author counts inflated; obscures real activity and quality trends.
- Recommendation: official filter batch accounts and publish dedup metrics; dashboard add “excl. anomalies” new count.
owner-cluster (severity: mid)
- Possible cause (hypothesis): account uckkk using templates to mass-generate plugins without real function.
- Impact: inflates total count, lowers avg score; users overwhelmed by low-quality content in search and rankings.
- Recommendation: mark uckkk as batch account and downrank; trigger manual review or removal.
dominant-owner (severity: low)
- Possible cause (hypothesis): single account massive registration, same source as owner-cluster but focusing on share impact.
- Impact: uckkk occupies 11% of ecosystem, distorts author and activity rankings, breaks fairness.
- Recommendation: set automated author share alert; single account exceeding threshold (e.g., 100 plugins) enters review queue.
npm-stale (severity: low)
- Possible cause (hypothesis): release pipeline not synced with repo, or authors not npm publish timely.
- Impact: users get outdated builds, may miss features or contain bugs; drift breaks dependency resolution.
- Recommendation: official add warning in market for repo/npm mismatch; authors configure auto-publish.
breaking-release (severity: high)
- Possible cause (hypothesis): official introducing API changes in alpha stage; dsh-v0.1.6-alpha.2 and alpha.1 both breaking.
- Impact: plugins depending on old interfaces may crash; especially those with manifest.no-client-export.
- Recommendation: official publish migration guide; authors check client export; users backup configs before upgrade.
Next-Stage Priorities & Recommendations
- Plugin authors:
- Run
npm publishimmediately to sync repo latest, remove npm-staleTop drift. - Check manifest and client entry against dsh-v0.1.6-alpha.2 changelog, fix no-client-export warnings.
- Add Chinese readme, topics, and demo GIFs to low-star high-quality plugins for discoverability.
- Avoid template batch, focus on single-point function with continuous maintenance.
- Run
- Ecosystem users:
- Prioritize grade S/A with weekly >1000, e.g., dsh-market, omdsh-dev/DSH-better-sidebar.
- Verify repo/npm version consistency before install to avoid outdated builds.
- Be cautious of high-download grade C plugins (e.g., genius-alray/dsh-mermaid-render); wait for official risk flags.
- Follow inAwesome/inImsai but also actively explore quadrant.gems high-quality low-star plugins.
- DSH official:
- Clean batch accounts uckkk, xby-skill; redefine scale as “dedup + quality-weighted”.
- Add “underrated gems” recommendation slots and “quality doubtful” warnings for low-score high-download plugins.
- Expand LLM capability tagging coverage (currently 1823/11792) for better theme map and search clustering.
- Publish migration guide for breaking release and set at least 2-week adaptation window.
Official Dynamics Interpretation
- dsh latest dsh-v0.1.6-alpha.2 (2026-09-17) breaking: plugin authors must adapt immediately, otherwise mass plugin failures after user upgrade.
- dshStars 231221, dshPushed 2026-09-17: platform itself hot, but plugin ecosystem not fully absorbing; many users not migrated to plugins.
- Model repos high popularity: DeepSeek-V3 (104472 stars), R1 (91984) etc.; plugin ecosystem should leverage model heat to drive installs.
- coverage authoritative 11792 < candidates 16294: over 4500 candidate plugins not included; official needs broaden inclusion to reflect reality.