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 1–20 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
🥇
HUB
5hackernews13

Establish a verified plugin marketplace

Define official qualification criteria and maintain a trusted, curated marketplace of valid DSH plugins. The marketplace should support accurate discovery, installation, updates, version tracking, and clear review and removal policies.

Explore challenge ↗

The problem

Define official qualification criteria and maintain a trusted, curated marketplace of valid DSH plugins. The marketplace should support accurate discovery, installation, updates, version tracking, and clear review and removal policies.

Possibly addressed by:Plugin Marketplace · Plugin Market · Plugin Market

Start with these discussions

  • 强烈建议尽快建立官方的插件市场,做好标准和安全规范,并有审核奖励等机制github · ▲80 · 25 comments
  • 官方应该出一个 DSH 插件市场,社区插件的检索功能基本已经废了github · ▲18 · 18 comments
  • 拿 DSH + V4 Flash + PTC 跑了一次真实 PR,几点体验和建议github · ▲4 · 4 comments
Last active Sep 12, 2026
9 authors9 posts · 106 reactions56 comments
🥈
PLUGIN
6hackernews4

Track model usage and enforce spending budgets

Record durable, accurately attributed usage and cost events for foreground, auxiliary, tool, compaction, retry, and subagent calls. Provide session totals, warnings, and configurable spending circuit breakers.

Explore challenge ↗

The problem

Record durable, accurately attributed usage and cost events for foreground, auxiliary, tool, compaction, retry, and subagent calls. Provide session totals, warnings, and configurable spending circuit breakers.

Possibly addressed by:Usage Cost · Token Usage Cost · Token Use

Start with these discussions

  • `tokenUsage` projection never folds `compaction/summary.usage`, and replaces rather than adds across a retried stepgithub · ▲4 · 28 comments
  • [Bug] Harness 界面 Token 统计严重偏低,实际账单 12 分钟扣费 50+ 元,请核查并优化github · ▲4 · 2 comments
🥉
HARNESS
52

Secure configurable remote binding

Keep listeners bound to localhost by default, while allowing explicit binding to a selected LAN interface or other address with authentication, host-trust protections, configurable access controls, clear exposure warnings, and gated wildcard binds such as `0.0.0.0`.

Explore challenge ↗

The problem

Keep listeners bound to localhost by default, while allowing explicit binding to a selected LAN interface or other address with authentication, host-trust protections, configurable access controls, clear exposure warnings, and gated wildcard binds such as `0.0.0.0`.

Possibly addressed by:Web Lan Access · Lan Proxy · Lan Gateway

Start with these discussions

  • 还没法--host 0.0.0.0启动啊github · ▲26 · 29 comments
  • 请问有计划增加认证层来放开0.0.0.0的监听吗
4
PLUGIN
11hackernews1

Configure per-model multimodal capabilities

Allow persistent per-model modality declarations and editable capability settings for text, image, audio, and video inputs across built-in, custom, and compatible providers. Document the configuration, report capability limits, and ensure attachments and multimodal plugin results use valid model routes.

Explore challenge ↗

The problem

Allow persistent per-model modality declarations and editable capability settings for text, image, audio, and video inputs across built-in, custom, and compatible providers. Document the configuration, report capability limits, and ensure attachments and multimodal plugin results use valid model routes.

Possibly addressed by:Model Modality · Image Vision · Models Input Modalities

Start with these discussions

  • BUG?:设置页无法声明模型的输入模态(图片),并导致图片无法正常输入github · ▲7 · 2 comments
  • 加入的模型无法设置是否支持图片!
5
HARNESS
6hackernews11

Ship an official cross-platform desktop client

Provide an officially supported, installable desktop application for macOS, Windows, and Linux that packages DSH, simplifies setup, and supports reliable updates and native integration.

Explore challenge ↗

The problem

Provide an officially supported, installable desktop application for macOS, Windows, and Linux that packages DSH, simplifies setup, and supports reliable updates and native integration.

Possibly addressed by:Desktop Shell · Desktop · Wx Desktop

Start with these discussions

  • 什么时候出官方桌面版?github · ▲3 · 6 comments
6
HARNESS
4

Configure per-conversation session-affinity headers

Attach stable per-conversation session identifiers, such as UUID-based OpenCode headers, to outbound inference requests. Allow provider- and model-level control over header names, formats, and omission, together with related prompt-cache settings to improve routing and cache affinity.

Explore challenge ↗

The problem

Attach stable per-conversation session identifiers, such as UUID-based OpenCode headers, to outbound inference requests. Allow provider- and model-level control over header names, formats, and omission, together with related prompt-cache settings to improve routing and cache affinity.

Possibly addressed by:Opencode Session Header · X Opencode Session · Session Header

Start with these discussions

  • Please send x-opencode-session header on API requests (OpenCode Go requires it from 09/05)github · ▲175 · 30 comments
  • bug(llm-pi-ai): openai-responses 路由无法关闭 session_id 请求头github · ▲1 · 2 comments
  • opencode-go 无法发送逐对话会话头(x-opencode-session):compat 门控 withhold 导致配置层无解,缓存命中率受损
7
HARNESS
8

Validate and isolate plugins at startup

Check plugin compatibility and registration during startup, report clear per-plugin diagnostics, and disable or quarantine failed third-party plugins while allowing healthy plugins to load. Preserve fail-fast behavior for core plugins.

Explore challenge ↗

The problem

Check plugin compatibility and registration during startup, report clear per-plugin diagnostics, and disable or quarantine failed third-party plugins while allowing healthy plugins to load. Preserve fail-fast behavior for core plugins.

Possibly addressed by:Plugin Security Review · Plugin Precheck · Plugin Check

Start with these discussions

  • [Bug] One third-party plugin load failure aborts the entire app, with no self-repair path / 单个第三方插件加载失败导致整个应用无法启动,且无自修复路径github · ▲3 · 2 comments
  • 现在的dsh和最开始的openclaw简直一模一样,一升级就容易挂github · ▲2 · 4 comments
  • Plugin loading lacks fault isolation: one failing plugin takes down the entire plugin tree
8
PLUGIN
11

Discover providers and models dynamically

Discover provider and model catalogs through live APIs, including third-party providers such as OpenRouter, and populate the settings UI with available models and metadata. Retain bundled, cached, or temporary override data as fallbacks when discovery is unavailable.

Explore challenge ↗

The problem

Discover provider and model catalogs through live APIs, including third-party providers such as OpenRouter, and populate the settings UI with available models and metadata. Retain bundled, cached, or temporary override data as fallbacks when discovery is unavailable.

Possibly addressed by:Model Catalog Refresh · Live Model Catalog · Catalog Refresh

Start with these discussions

  • Provider model catalogs are stale for OpenCode Go and OpenRoutergithub · ▲11 · 4 comments
  • 支持 OpenCode Go 的最新模型列表github · ▲6 · 1 comments
  • 针对第三方提供商配置体验优化的问题github · ▲2 · 1 comments
9
PLUGIN
6

Configure SSRF-Safe Fake-IP Resolution Allowlisting

Allow web-fetch plugins to configure trusted resolvers, domains, or CIDR ranges for transparent-proxy fake-IP DNS while retaining SSRF protection. Support relevant IPv4 and IPv6 pools, keep private and link-local networks blocked by default, and provide actionable errors.

Explore challenge ↗

The problem

Allow web-fetch plugins to configure trusted resolvers, domains, or CIDR ranges for transparent-proxy fake-IP DNS while retaining SSRF protection. Support relevant IPv4 and IPv6 pools, keep private and link-local networks blocked by default, and provide actionable errors.

Possibly addressed by:Web Fetch Enhanced · Fakeip Compat · Fetch Third Party

Start with these discussions

  • web-fetch-http: add a config toggle to allow non-public DNS answers (fake-ip proxy compatibility)github · ▲12 · 4 comments
  • 建议为 web_fetch 增加可配置的非公网 CIDR 白名单,以支持 Fake-IP 与特殊路由环境github · ▲6 · 1 comments
  • web-fetch-http SSRF guard rejects every fetch under transparent-proxy fake-ip DNS (198.18.0.0/15) - no resolver/allowlist escape hatch
10
PLUGIN
41

Configure Per-Request Timeouts

Support configurable longer timeouts for large-payload RPCs and model requests, including slow local model loading, generation, retries, and large outputs, while retaining shorter defaults for ordinary calls.

Explore challenge ↗

The problem

Support configurable longer timeouts for large-payload RPCs and model requests, including slow local model loading, generation, retries, and large outputs, while retaining shorter defaults for ordinary calls.

Start with these discussions

  • llm-pi-ai Timeout increase for local LLMsgithub · ▲14 · 18 comments
  • 使用本地部署的模型时,如果重启过电脑,Vram内容被清空,则发起对话时模型要思考很久时间,如果超过五分钟就报错重试了,导致无法继续对话github · ▲1 · 2 comments
  • dsh-client-connection:history 等大载荷 RPC 受全局 30s 超时限制,Web 端报「历史加载失败:signal timed out (internal)」—— timeoutPolicy 钩子已存在但无人接线github · ▲1 · 1 comments
11
HARNESS
8

Make Web UI authentication configurable

Add configurable authentication for the Web UI across local and remote deployments, including user tokens and external providers such as OAuth. Preserve secure defaults while allowing users to explicitly disable the built-in one-time token when appropriate.

Explore challenge ↗

The problem

Add configurable authentication for the Web UI across local and remote deployments, including user tokens and external providers such as OAuth. Preserve secure defaults while allowing users to explicitly disable the built-in one-time token when appropriate.

Possibly addressed by:Web Auth · Auth Proxy · Web Auth

Start with these discussions

  • Allow bypass one-time token authenticationgithub · ▲5 · 2 comments
  • 访问web的URL新增token认证不方便github · ▲3 · 1 comments
  • dsh web: provide an opt-out for the mandatory browser token auth (reverse-proxy / trusted-LAN deployments)github · ▲2 · 1 comments
12
HARNESS
6

Support multi-folder workspaces

Allow a workspace to contain multiple ordered directories with an optional primary directory while preserving single-folder compatibility. Sessions, configuration discovery, Git operations, and sandbox permissions should coordinate across all associated directories.

Explore challenge ↗

The problem

Allow a workspace to contain multiple ordered directories with an optional primary directory while preserving single-folder compatibility. Sessions, configuration discovery, Git operations, and sandbox permissions should coordinate across all associated directories.

Possibly addressed by:Multi Workspace · Workspace Merge · Multi Folder

Start with these discussions

  • Feature: Workspace should support multiple root paths (project with appended directories)github · ▲4 · 4 comments
  • [Feature Idea] 多文件夹工作区与 primary 文件夹(对齐 Codex 的项目模型)github · ▲3 · 1 comments
  • 工作区能添加多个文件夹目录吗?
13
HARNESS
6hackernews2

Limit runaway deliberation and request clarification

Add safeguards that detect excessive, repetitive, ambiguous, or stalled agent behavior, cap or halt it before wasting resources, and ask the user for clarification when intent or the next action is unclear. Support recovery or strategy changes where possible.

Explore challenge ↗

The problem

Add safeguards that detect excessive, repetitive, ambiguous, or stalled agent behavior, cap or halt it before wasting resources, and ask the user for clarification when intent or the next action is unclear. Support recovery or strategy changes where possible.

Possibly addressed by:Circuit Breaker · Continue On Limit · Loop Continue

Start with these discussions

  • agent 在长任务中回复陷入循环、大量重复,且无轮次预算/重复检测兜底github · ▲3 · 8 comments
  • [Bug] Agent output enters an infinite repetition loop that runs for ~10 minutes with no circuit breaker — only a manual abort stops it
14
PLUGIN
4hackernews6

Strengthen Agent Sandbox Isolation

Run agents in rootless, strongly isolated sandboxes with fine-grained filesystem, command, privilege, resource, network, X11, and D-Bus controls. Restrict access to approved workspaces and data, protect credentials such as Xauthority, and prevent shell or socket-based bypasses and sensitive-data exfiltration.

Explore challenge ↗

The problem

Run agents in rootless, strongly isolated sandboxes with fine-grained filesystem, command, privilege, resource, network, X11, and D-Bus controls. Restrict access to approved workspaces and data, protect credentials such as Xauthority, and prevent shell or socket-based bypasses and sensitive-data exfiltration.

Possibly addressed by:Sandbox Micro · Agentenv Sandbox · Sandbox Allowlist

Start with these discussions

  • Hey! Someone should check wasmer sandboxes works on web so... could fit for this harness?github · ▲2 · 1 comments
  • Support rootless container toolchains under `workspace-write`
15
HARNESS
5

Support Backward-Compatible Persistence Migration

Add versioned, tolerant migration and backward-compatible reads for legacy logs, sessions, and durable data, including safe handling, repair, or degradation of unknown ignorable events and obsolete plugin fields without data loss. Document the supported channel for UI-only notifications that should not be persisted.

Explore challenge ↗

The problem

Add versioned, tolerant migration and backward-compatible reads for legacy logs, sessions, and durable data, including safe handling, repair, or degradation of unknown ignorable events and obsolete plugin fields without data loss. Document the supported channel for UI-only notifications that should not be persisted.

Possibly addressed by:Session Migration Repair · Session Migrate · Session Migrator

Start with these discussions

  • 升级到 0.1.5 后,以前的一些老会话打不开了github · ▲5 · 6 comments
  • v0→v3 session migration refuses legacy plugin-injected message sources (history fails to load after host upgrade)github · ▲2 · 2 comments
16
HARNESS
7

Register Plugin Session Events

Let plugins declare custom persisted or durable session event types so the harness recognizes them when loading and re-observing sessions, while safely skipping declared ignorable events for older readers.

Explore challenge ↗

The problem

Let plugins declare custom persisted or durable session event types so the harness recognizes them when loading and re-observing sessions, while safely skipping declared ignorable events for older readers.

Start with these discussions

  • `Session.append` cannot set `ignorable`, so a plugin's own event type makes its sessions unloadablegithub · ▲3 · 2 comments
  • [Proposal] Session event registration for out-of-tree plugins — unknown event types refuse resume (dsh-click/observed|action crashes rc.6/rc.7)github · ▲2 · 13 comments
  • Out-of-repo plugins cannot mark session events `ignorable` — every log they touch becomes unreadable (extensibility gap + minimal fix)github · ▲2 · 3 comments
Last active Sep 9, 2026
7 authors7 posts · 12 reactions23 comments
17
HARNESS
2hackernews32

Connect subscription providers via OAuth

Provide an authorized OAuth flow that lets DSH and other harnesses use existing Claude or other provider subscriptions without API keys, with clear authorization, status, and error handling to reduce account-risk concerns.

Explore challenge ↗

The problem

Provide an authorized OAuth flow that lets DSH and other harnesses use existing Claude or other provider subscriptions without API keys, with clear authorization, status, and error handling to reduce account-risk concerns.

Possibly addressed by:Codex Oauth · Openai Codex Auth · Subscriptions

Start with these discussions

  • Support signing in to subscription-only LLM providers (OpenAI Codex)
18
HARNESS
4hackernews5

Configure models and reasoning per teammate

Allow team leads or agent-teams users to independently select a provider, model, and reasoning effort for each teammate, including cheaper or local models.

Explore challenge ↗

The problem

Allow team leads or agent-teams users to independently select a provider, model, and reasoning effort for each teammate, including cheaper or local models.

Possibly addressed by:Agent Team Model · Agent Teams · Agent Teams

Start with these discussions

  • tool-agent-team:agent/created 时序竞态会随机杀死一次性子代理(TEAM_NOT_MEMBER)github · ▲1 · 2 comments
  • Model selection in agent-teamsgithub · ▲1 · 0 comments
  • 【功能建议】Agent Team模式建议加入可以让Lead给teammate指定模型的功能
19
PLUGIN
6

Configure per-model reasoning effort

Allow reasoning or thinking intensity to be configured per model and route, including locally hosted llama-server models and Codex subagents. Support defined levels, defaults, mappings, disabled reasoning, and clear handling of invalid values.

Explore challenge ↗

The problem

Allow reasoning or thinking intensity to be configured per model and route, including locally hosted llama-server models and Codex subagents. Support defined levels, defaults, mappings, disabled reasoning, and clear handling of invalid values.

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

Start with these discussions

  • 使用llama-server本地部署的deepseek v4 flash 如何在deepseek harness中调整思考强度?github · ▲1 · 6 comments
  • 添加自定义模型无法设置思考模式github · ▲1 · 4 comments
  • subagent-codex 后端支持配置 reasoning effort // configurable reasoning effort for the Codex subagent backend
20
HARNESS
6

Delete sessions permanently

Add a confirmed action to delete active or archived sessions and remove their associated persisted and in-memory data. Emit a typed deletion lifecycle event so plugins can respond to the removal.

Explore challenge ↗

The problem

Add a confirmed action to delete active or archived sessions and remove their associated persisted and in-memory data. Emit a typed deletion lifecycle event so plugins can respond to the removal.

Possibly addressed by:Session Admin · Session Delete · Delete Session

Start with these discussions

  • Feature Request: Support native /delete command to permanently delete sessionsgithub · ▲4 · 2 comments
  • add delete buttongithub · ▲2 · 2 comments
  • SessionPersistence.delete: a single-session deletion primitive (proposal with a complete, tested implementation)github · ▲1 · 3 comments
Last active Sep 12, 2026
← PreviousPage 1 of 9Next →
[反馈] web_search 的两个成本相关点:调用 token 未计入会话用量、单次调用成本偏高
github · ▲2 · 1 comments
Last active Sep 19, 2026
10 authors10 posts · 13 reactions38 comments
github · ▲3 · 2 comments
  • 为什么要一刀切禁止监听0.0.0.0?github · ▲2 · 5 comments
  • Last active Sep 18, 2026
    6 authors7 posts · 34 reactions37 comments
    github · ▲3 · 6 comments
  • 给模型设置加个多模态支持吧,每次改配置文件不方便。github · ▲3 · 1 comments
  • Last active Sep 15, 2026
    12 authors12 posts · 21 reactions21 comments
    抓紧出官方的桌面版。
    github · ▲3 · 2 comments
  • DeepSeek Harness really needs an official desktop clientgithub · ▲3 · 0 comments
  • Last active Sep 20, 2026
    8 authors8 posts · 15 reactions8 comments
    github · ▲1 · 1 comments
    Last active Sep 10, 2026
    4 authors4 posts · 178 reactions34 comments
    github · ▲2 · 3 comments
    Last active Sep 18, 2026
    8 authors8 posts · 12 reactions18 comments
    Last active Sep 12, 2026
    10 authors11 posts · 28 reactions11 comments
    github · ▲3 · 10 comments
    Last active Sep 15, 2026
    6 authors6 posts · 25 reactions16 comments
    Last active Sep 19, 2026
    4 authors5 posts · 17 reactions22 comments
    Last active Sep 17, 2026
    8 authors8 posts · 15 reactions5 comments
    github · ▲2 · 1 comments
    Last active Sep 19, 2026
    6 authors6 posts · 12 reactions7 comments
    github · ▲1 · 3 comments
  • 大模型过度思考github · ▲1 · 1 comments
  • Last active Sep 16, 2026
    8 authors8 posts · 8 reactions14 comments
    github · ▲1 · 1 comments
  • bwrap sandbox can be bypassed by accessing X11 socket or dbusgithub · ▲1 · 0 comments
  • Last active Sep 19, 2026
    10 authors10 posts · 5 reactions2 comments
    0.1.5-rc.1: v0→v3 迁移有三道 fail-closed 闸门 —— 49/123 之前的会话点不开(含没人报过的 inbox/spliced 一类),附已验证的修复配方 0/49→49/49
    github · ▲1 · 11 comments
    Last active Sep 14, 2026
    5 authors5 posts · 10 reactions22 comments
    github · ▲2 · 1 comments
  • 把 ChatGPT / Codex 订阅直接接入 DeepSeek Harness,用已有的 ChatGPT 账号在 DSH 里用 Codex 模型,不需要 OpenAI API Key 也不用装 Codex CLIx · ▲2 · 0 comments
  • 内置 openai-codex能在 0.1.2-rc.1 中完成 OAuth 登录吗?github · ▲1 · 2 comments
  • Last active Sep 18, 2026
    7 authors7 posts · 5 reactions3 comments
    github · ▲1 · 0 comments
    Last active Sep 17, 2026
    9 authors9 posts · 4 reactions2 comments
    github · ▲1 · 3 comments
    Last active Sep 14, 2026
    6 authors6 posts · 6 reactions15 comments
    6 authors6 posts · 10 reactions7 comments