DeepSeek Harness Plugin Hub

Publish and manage complete Harness Profiles. Discover Plugins for your next setup.

Explore

PluginsPresetsDocsNews

Community

Publish a pluginContactReport an issue

Resources

Plugin Hub on GitHubDeepSeek HarnessSystem statusPrivacy notice
© 2026 DeepSeek Harness Plugin HubPowered byPaxTech

Independent and unofficial. Not affiliated with, authorized by, or endorsed by DeepSeek.

Community Feature Requests — Plugin Hub News
DeepSeek Harness Plugin Hub
ProfilesPluginsCategoriesNewsDocsSign inManage Profiles
ProfilesPluginsCategoriesNewsDocsSign in

COMMUNITY CHALLENGES

Find your next challenge.

Real problems from the community. Find one that sparks an idea, explore the discussion, and build something people need.

Showing 61–80 of 164 community requestsUpdated hourly from community discussions
All updatesDSH ReleasesFeature Requests
All3 days7 days14 days
TopNew
OpenShippedSolved by pluginWon't fixAll statuses
#Problems worth solvingCommunity demand
61
HARNESS
hackernews6

Automatically discover and apply instruction files

Automatically discover and apply shared, harness-specific, and nested AGENTS.md instruction files, including directory-specific guidance.

Explore challenge ↗

The problem

Automatically discover and apply shared, harness-specific, and nested AGENTS.md instruction files, including directory-specific guidance.

Possibly addressed by:Agents Md Manager · Agent Instructions Plus · Custom Instructions

Start with these discussions

  • Comment on: Claude Code now reads AGENTS.md if there is no Claude.mdhackernews · ▲0 · 0 comments
  • Comment on: Claude Code now reads AGENTS.md if there is no Claude.mdhackernews · ▲0 · 0 comments
  • Comment on: Claude Code now reads AGENTS.md if there is no Claude.mdhackernews · ▲0 · 0 comments
Last active Sep 19, 2026
6 authors6 posts · 0 reactions0 comments
62
HARNESS
1hackernews2

Enforce user and project instructions

Make the agent reliably follow the user's intent and applicable project instruction files such as AGENTS.md, while respecting higher-priority system, developer, and direct user instructions when conflicts occur.

Explore challenge ↗

The problem

Make the agent reliably follow the user's intent and applicable project instruction files such as AGENTS.md, while respecting higher-priority system, developer, and direct user instructions when conflicts occur.

Possibly addressed by:Agent Instructions As Prompt · Agent Instructions Plus · Custom Instructions

Start with these discussions

  • 所有注入的 AGENTS.md 指令都被"仅供参考"式前言弱化，建议改为约束性表述github · ▲2 · 0 comments
  • Comment on: Claude Code now reads AGENTS.md if there is no Claude.mdhackernews · ▲0 · 0 comments
63
HARNESS
2

Configure LAN and reverse-proxy access

Provide documented, persistent configuration for using DSH from other LAN devices or through a reverse proxy, including host, proxy, port, origin, and related access settings. The configuration should support NAS and other remote deployments without source changes or fragile workarounds.

Explore challenge ↗

The problem

Provide documented, persistent configuration for using DSH from other LAN devices or through a reverse proxy, including host, proxy, port, origin, and related access settings. The configuration should support NAS and other remote deployments without source changes or fragile workarounds.

Possibly addressed by:Lan Proxy · Plugin Lan Proxy · Lan Proxy

Start with these discussions

  • 为什么DSH对LAN口'0.0.0.0'的监听要做如此严格的限制github · ▲1 · 2 comments
  • 能不能开放一下DSH局域网内跨设备访问啊？github · ▲1 · 1 comments
Last active Sep 18, 2026
2 authors2 posts · 2 reactions3 comments
64
HARNESS
3hackernews1

Stabilize and document the extension API

Provide an officially supported extension directory, lifecycle, loading mechanism, and documented stable plugin contract that remains compatible across refactors and harness versions. The API should cover service injection, route and remote method registration, prompt and tool integration, schema handling, reload behavior, public exports, and persistent third-party plugins.

Explore challenge ↗

The problem

Provide an officially supported extension directory, lifecycle, loading mechanism, and documented stable plugin contract that remains compatible across refactors and harness versions. The API should cover service injection, route and remote method registration, prompt and tool integration, schema handling, reload behavior, public exports, and persistent third-party plugins.

Possibly addressed by:Plugin Kit · Plugin Dev Manager · Plugin Market

Start with these discussions

  • Feature Request: Official extension directory for user plugins with stable service injectiongithub · ▲1 · 1 comments
65
HARNESS
2hackernews1

Add per-subagent model configuration

Let users configure a dedicated provider and model for child agents through a clear subagent control UI, while preserving explicit per-agent overrides and inheriting the parent model when no dedicated configuration is set.

Explore challenge ↗

The problem

Let users configure a dedicated provider and model for child agents through a clear subagent control UI, while preserving explicit per-agent overrides and inheriting the parent model when no dedicated configuration is set.

Possibly addressed by:Subagent Model Picker · Submodel Change · Subagent Model

Start with these discussions

  • 每次切换主模型的时候，子agent模型会跟随主模型一起变更（同主模型），怎么才能固定子agent模型？github · ▲2 · 3 comments
  • Feature request: 子代理模型可配置（agent-child-model 设置分节）github · ▲1 · 0 comments
66
HARNESS
2

Replay Session History on ACP Load

Add ACP session/load support and replay persisted session updates when loading or resuming a session. This lets clients rebuild the full conversation transcript after restart instead of receiving only the resumed state and future updates.

Explore challenge ↗

The problem

Add ACP session/load support and replay persisted session updates when loading or resuming a session. This lets clients rebuild the full conversation transcript after restart instead of receiving only the resumed state and future updates.

Possibly addressed by:Acp Replay · Autoload History · Session Rewind

Start with these discussions

  • [ACP] session/resume restores the agent but does not replay history; loadSession is missing, so ACP clients show a blank transcriptgithub · ▲2 · 3 comments
  • [dsh-acp] ACP clients cannot rebuild a session transcript: no session/load, and session/resume does not replay historygithub · ▲1 · 2 comments
Last active Sep 12, 2026
2
67
PLUGIN
2

Enforce configurable filesystem root jails

Restrict agents and plugins to configured per-profile or workspace roots, rejecting paths outside them unless explicitly authorized. Preserve unrestricted access when no root is configured and clearly report files that are unavailable within the allowed workspace.

Explore challenge ↗

The problem

Restrict agents and plugins to configured per-profile or workspace roots, rejecting paths outside them unless explicitly authorized. Preserve unrestricted access when no root is configured and clearly report files that are unavailable within the allowed workspace.

Possibly addressed by:Fs Deny Policy · Workspace Envrc · Remote Workspace

Start with these discussions

  • sandbox：网页版agent会访问工作区外的文件夹？github · ▲1 · 2 comments
  • [Idea] Optional root jail for the fs service (per-profile workspace root) — safer multi-agent setupsgithub · ▲1 · 0 comments
Last active Sep 18, 2026
2 authors2 posts · 2 reactions
68
HARNESS
2

Select Model and Effort Per Session

Allow each session or launch to choose its own model and effort level without changing the deployment-wide default. Propagate these settings through session creation and agent resolution.

Explore challenge ↗

The problem

Allow each session or launch to choose its own model and effort level without changing the deployment-wide default. Propagate these settings through session creation and agent resolution.

Possibly addressed by:Subagent Model Picker · Fixed New Session Model · Submodel Change

Start with these discussions

  • Session-only model selection for Agent Sessions lanesgithub · ▲2 · 1 comments
  • Session-scoped model and effort selectiongithub · ▲2 · 1 comments
Last active Sep 19, 2026
1 authors2 posts · 4 reactions2 comments
69
HARNESS
2

Add expandable multiline goal display and editing

Make the GoalBar expandable so long objectives can be fully read, provide a full-text hover fallback, and replace the single-line editor with a bounded multiline textarea. Preserve the compact collapsed layout and existing save, cancel, and empty-goal validation behavior.

Explore challenge ↗

The problem

Make the GoalBar expandable so long objectives can be fully read, provide a full-text hover fallback, and replace the single-line editor with a bounded multiline textarea. Preserve the compact collapsed layout and existing save, cancel, and empty-goal validation behavior.

Possibly addressed by:Goal Mode · Grok Goals · Goal

Start with these discussions

  • Proposal: support multiline editing for active goal objectivesgithub · ▲1 · 2 comments
  • [功能建议] 目标栏只有一行且无法展开：长目标读不全，内联编辑框也只有单行github · ▲1 · 0 comments
Last active Sep 18, 2026
2 authors2 posts · 2 reactions2 comments
70
HARNESS
2

Add universal automatic RTL support to the web UI

Add automatic right-to-left direction detection across the chat composer, user messages, and assistant markdown responses so Hebrew, Arabic, Persian, Urdu, and mixed RTL/LTR content render correctly. The capability should be integrated into the core UI rather than requiring a manual patch and installer.

Explore challenge ↗

The problem

Add automatic right-to-left direction detection across the chat composer, user messages, and assistant markdown responses so Hebrew, Arabic, Persian, Urdu, and mixed RTL/LTR content render correctly. The capability should be integrated into the core UI rather than requiring a manual patch and installer.

Possibly addressed by:Client Ui Rtl · Hebrew Rtl · Chat Index Rail

Start with these discussions

  • [Feature & Fix] - Universal RTL Support for Hebrew, Arabic & Right-to-Left Languages (Ready Patch & Installer) ❤️github · ▲3 · 2 comments
  • Support RTL rendering for Arabic text in the chat interfacegithub · ▲1 · 0 comments
Last active Sep 12, 2026
71
HARNESS
2

Show repository status beside workspace names

Display each workspace’s active branch and Git state in the sidebar, including clean or dirty status and ahead/behind information, with detailed repository data available on hover.

Explore challenge ↗

The problem

Display each workspace’s active branch and Git state in the sidebar, including clean or dirty status and ahead/behind information, with detailed repository data available on hover.

Possibly addressed by:Git Branch · Workspace Overview · Workspace Status Badge

Start with these discussions

  • Feature request: Git status badge on sidebar workspace rowsgithub · ▲1 · 2 comments
  • Show repository branch name along side workspace name on the sidebargithub · ▲1 · 0 comments
Last active Sep 18, 2026
2 authors2 posts · 2 reactions2 comments
72
HARNESS
2

Support WebMCP Tool Discovery and Reuse

Enable DSH agents and browser plugins to discover and invoke WebMCP tools, then build, verify, version, manage, and reuse them across websites. This includes the core WebMCP support requested for using page-provided capabilities while reducing repeated interactions and token costs.

Explore challenge ↗

The problem

Enable DSH agents and browser plugins to discover and invoke WebMCP tools, then build, verify, version, manage, and reuse them across websites. This includes the core WebMCP support requested for using page-provided capabilities while reducing repeated interactions and token costs.

Possibly addressed by:Web Bridge Mcp · Mcphub · Mcp Manager

Start with these discussions

  • Support webMCPgithub · ▲3 · 3 comments
  • DeepSeek Harness + WebMCP + Builder = Efficient Computer Usegithub · ▲2 · 2 comments
Last active Sep 7, 2026
2 authors2 posts · 5 reactions
73
PLUGIN
2

Support configuring additional sandbox writable roots

Allow sandbox policy configuration to specify additional writable root directories, such as SDK installation paths, so toolchain caches can be written without expanding the workspace or requiring repeated permission escalation.

Explore challenge ↗

The problem

Allow sandbox policy configuration to specify additional writable root directories, such as SDK installation paths, so toolchain caches can be written without expanding the workspace or requiring repeated permission escalation.

Possibly addressed by:Write Protect · Sandbox Allowlist · Plugin Multi Root Workspace

Start with these discussions

  • Sandbox writableRoots hardcoded — no way to add extra writable dirs for SDK toolchains / 沙箱可写根目录无法附加(SDK工具链)github · ▲2 · 1 comments
  • [Feature request] workspace-write 增加 additionalWritableRoots：支持工作区外共享工具链缓存的可写配置github · ▲1 · 2 comments
Last active Sep 13, 2026
74
GENERAL
2

Expose per-model reasoning effort metadata

Expose each model’s supported reasoning-effort values and default, including custom-provider models, through model metadata or configuration. Reject unsupported values instead of silently ignoring them so clients can build accurate selectors.

Explore challenge ↗

The problem

Expose each model’s supported reasoning-effort values and default, including custom-provider models, through model metadata or configuration. Reject unsupported values instead of silently ignoring them so clients can build accurate selectors.

Possibly addressed by:Reasoning Effort · More Models Thinking Level · Better Reasoning Effort

Start with these discussions

  • 解决一下这个,自定义提供商默认不能设置思考等级github · ▲3 · 1 comments
  • Expose reasoning effort metadata in `GET /models`github · ▲1 · 0 comments
Last active Sep 13, 2026
2 authors2 posts · 4 reactions1 comments
75
HARNESS
2

Optimize long-session history storage and loading

Clean up redundant finalized streaming chunks while preserving semantic messages, final responses, tool calls, and results, and load long histories lazily or paginated to avoid excessive memory use, stack overflows, and browser freezes.

Explore challenge ↗

The problem

Clean up redundant finalized streaming chunks while preserving semantic messages, final responses, tool calls, and results, and load long histories lazily or paginated to avoid excessive memory use, stack overflows, and browser freezes.

Possibly addressed by:Session Handoff · Session Handoff · Plugin Session Archive

Start with these discussions

  • [bug反馈] 超长会话因流式增量碎片（reasoning-chunks / text-chunks）未自动清理导致内存溢出 (OOM) 与历史加载失败github · ▲8 · 2 comments
  • Web版无法打开超长会话，提示 Maximum call stack size exceededgithub · ▲2 · 2 comments
Last active Sep 1, 2026
2 authors2 posts · 10 reactions
76
HARNESS
2

Keep the web UI responsive during concurrent subagent streaming

Improve responsiveness and reliability of the web interface when many subagents run concurrently by batching streamed events, coalescing token updates, and avoiding excessive client-side rendering overhead.

Explore challenge ↗

The problem

Improve responsiveness and reliability of the web interface when many subagents run concurrently by batching streamed events, coalescing token updates, and avoiding excessive client-side rendering overhead.

Possibly addressed by:Subagent Team · Subagents · Background Agents

Start with these discussions

  • 当并发多个subagent时，web界面加载会变慢，甚至加载不出，或许可以为每个subagent分配一个线程以解决这个问题github · ▲5 · 1 comments
  • UI lag during token streaming / many background subagents: measured root cause (per-token fan-out, no stream batching)github · ▲1 · 2 comments
Last active Sep 6, 2026
2 authors2 posts · 6 reactions3 comments
77
HARNESS
2

Detect and stop repetitive generation loops

Detect excessive repetition in generated content, including reasoning blocks and n-gram or near-identical text patterns, then interrupt the degraded generation and notify the user. The guard should avoid false positives on legitimate long reasoning where possible.

Explore challenge ↗

The problem

Detect excessive repetition in generated content, including reasoning blocks and n-gram or near-identical text patterns, then interrupt the degraded generation and notify the user. The guard should avoid false positives on legitimate long reasoning where possible.

Possibly addressed by:Repetition Guard · Loop Guard · Dupguard

Start with these discussions

  • 长会话下输出陷入重复循环（degenerate repetition），建议加输出重复检测github · ▲1 · 1 comments
  • Runaway repetition in the reasoning channel: one 736 KB block with 48,898 repeated phrases, and no guard to stop itgithub · ▲1 · 0 comments
Last active Sep 18, 2026
2 authors2 posts · 2 reactions
78
HARNESS
hackernews41

Enable provider-agnostic multi-model workflows

Support arbitrary models and providers, including local and frontier models, and allow workflows to delegate across them or switch between them without restarting the harness.

Explore challenge ↗

The problem

Support arbitrary models and providers, including local and frontier models, and allow workflows to delegate across them or switch between them without restarting the harness.

Possibly addressed by:Better Model Provider · Model Sync · Model Provider

Start with these discussions

  • Comment on: Claude Code now reads AGENTS.md if there is no Claude.mdhackernews · ▲0 · 0 comments
  • DeepSeek Harness:x · ▲0 · 0 comments
  • Comment on: The Harness Is the Thinghackernews · ▲0 · 0 comments
Last active Sep 18, 2026
5 authors5 posts · 0 reactions0 comments
79
HARNESS
2

Add lazy loading for session lists

Make workspace session lists load lazily or incrementally so large lists remain usable over slow or tunneled network connections and avoid timeouts.

Explore challenge ↗

The problem

Make workspace session lists load lazily or incrementally so large lists remain usable over slow or tunneled network connections and avoid timeouts.

Possibly addressed by:More Session Operations · Session Usage · Session Xc

Start with these discussions

  • 远程打开dsh web页面,无法打开session 列表github · ▲1 · 1 comments
  • [bug] `@deepseek-ai/dsh-api-session-controller` 的 `session/list` 默认全量加载，随使用数据逐渐爆炸，建议增加分页、懒加载机制github · ▲1 · 0 comments
Last active Sep 17, 2026
2 authors2 posts · 2 reactions1 comments
80
HARNESS
2

Make the Plan View Persistent and Expandable

Keep the plan view available after selections and provide a collapsible, resizable, or expanded layout for comfortable repeated viewing. Preserve plan actions such as approving the plan and continuing to plan.

Explore challenge ↗

The problem

Keep the plan view available after selections and provide a collapsible, resizable, or expanded layout for comfortable repeated viewing. Preserve plan actions such as approving the plan and continuing to plan.

Possibly addressed by:Plan Plus · Plugin Better Plan · Plan Toggle

Start with these discussions

  • [功能建议] plan 模式的计划视图增加全屏 / 侧边栏查看能力github · ▲3 · 1 comments
  • 计划模式中的计划面板能否支持反复查看？github · ▲1 · 0 comments
Last active Sep 11, 2026
2 authors2 posts · 4 reactions1 comments
← PreviousPage 4 of 9Next →
Comment on: AI agents lie, cheat and steal. That is putting off users
hackernews · ▲0 · 0 comments
Last active Sep 19, 2026
3 authors3 posts · 2 reactions0 comments
  • [Idea] 文件级自学习回路：把完成的会话蒸馏成跨会话耐久教训（附纯插件实证 @garvel/dsh-reflect）github · ▲1 · 0 comments
  • Plugin API gap on master: settingsNamespace / registerContinuableSetup / deepFreeze missinggithub · ▲1 · 0 comments
    • Last active Sep 8, 2026
    4 authors4 posts · 3 reactions1 comments
    Comment on: The Harness Is the Thing
    hackernews · ▲0 · 0 comments
    Last active Sep 9, 2026
    3 authors3 posts · 3 reactions3 comments
    authors
    2 posts · 3 reactions
    5 comments
    2 comments
    2 authors2 posts · 4 reactions2 comments
    5 comments
    2 authors
    2 posts · 3 reactions
    3 comments
    4 comments
    1 comments