dsh-hud: DSH Web Floating Status Panel
dsh-hud is a community HUD plugin for the DeepSeek Harness Web interface: the input bar dashboard button opens a right-side floating panel, centrally displaying Git status and diffs, MCP services, Skills, model and token usage, session statistics, and account balance. Installable via `dsh plugin --profile web`, it is read-only and requires no build, making it suitable for quickly confirming repository and session status within the chat interface.
Read Moredsh-plugin-garmin-connect: Integrating Garmin Connect Sports and Health Data into the DSH Agent
dsh-plugin-garmin-connect (v0.1.5, maintained by Likenttt) is a Garmin Connect integration plugin for DeepSeek Harness. It registers activities, sleep, steps, heart rate, weight, training libraries, and other data as 10 AI-callable tools, supporting FIT downloads and training creation. Installed via `npx @deepseek-ai/dsh plugin add`, it allows natural language queries in web dialogs or standalone MCP clients after configuring the GARMIN environment variable. Note that version 0.1.5 does not yet support browser-based two-factor authentication; the plugin runs with dsh process permissions, so review the source code and MIT license before installation.
Read Moredsh-harness-mcp-server: Exposing Harness Capabilities as an MCP Service
dsh-harness-mcp-server is a DSH workflow plugin released by community maintainer chushixixin, which exposes the DeepSeek Harness agent runtime as a StreamableHTTP MCP service. External MCP clients (such as Hermes) can drive the Harness to execute real coding tasks like bash commands and file system operations through tools like agent_run and task_inbox, and retrieve structured results including changes and verification. The plugin listens on 127.0.0.1:8090 by default, supports npm installation and activation via cordis.yml patches, and is suitable for alternative scenarios requiring context isolation or parallel execution.
Read Moredsh-oomol: Progressive MCP Discovery for Invoking OOMOL Connector Actions in DeepSeek Harness
dsh-oomol is a DeepSeek Harness web connectivity tool plugin maintained by oomol-lab, which connects the Harness to the OOMOL Connector (either OOMOL Hosted or self-hosted OpenConnector). The Harness only stores the Connector client keys, while Provider credentials remain on the Connector side; Actions are exposed on-demand through progressive MCP discovery. This article introduces the installation commands, dual-mode configuration, typical conversational usage, and security boundaries, along with applicable scenarios and troubleshooting tips.
Read Moredsh-mcp-manager: Visually Managing MCP Servers in DSH
dsh-mcp-manager is a client plugin developed by Js2Hou for DeepSeek Harness, which lists installed MCP servers in the "Settings → MCP" section. It supports adding, removing, enabling/disabling, editing, and connection testing for MCP servers, while displaying real-time connection status and the number of tools. Configuration changes are written back to cordis.patch.yml via HMR, eliminating the need for manual YAML editing. The plugin can be installed using the command `dsh plugin --profile web add @js2hou/dsh-mcp-manager` or through community marketplaces such as dsh-market and DSH Desktop. Licensed under MIT, it is ideal for DSH users who need centralized management of multiple MCP servers.
Read Moredsh-chatgpt-bridge: Create and Supervise DSH Sessions with ChatGPT Web
dsh-chatgpt-bridge is an independent DSH plugin (v0.4.1, MIT) maintained by jiezeng2004-design. It enables ChatGPT Web to connect to the local bridge via MCP through a secure tunnel (default 127.0.0.1:3456/mcp), allowing the creation, viewing, continuation, and supervision of DSH agent sessions. It exposes 15 tools, including Goal supervision and approval. The installation command is `dsh plugin --profile web add dsh-chatgpt-bridge`, requiring Node.js >= 22 and sharing the web profile with `dsh web`. The control flow is from ChatGPT to DSH, without altering the DSH core security model.
Read Moredsh-mcp-manager: A MCP Management Plugin for DSH with OAuth and stdio Capabilities
dsh-mcp-manager is a DSH web plugin maintained by hyqhyq3, managing MCP servers on the Settings → MCP page. It supports HTTP OAuth (PKCE + dynamic client registration), static tokens (environment variable names not stored on disk), and stdio local processes, with tools registered under the mcp__<name>__* namespace. Optional features include on-demand proxy schema compression, support for workspace-level mcp.json, and exclude blocking. Installation: npx -p @deepseek-ai/dsh dsh plugin --profile web add github:hyqhyq3/dsh-mcp-manager.
Read Moredsh-search-mcp: Using Search MCP to Replace DSH's Built-in Web Search
dsh-search-mcp is a DSH networking tool plugin maintained by gxpppp. While retaining the web_search tool name and display on the model side, it redirects search requests to Tavily, Brave, Exa, Perplexity, DuckDuckGo, or custom MCP servers, and disables the built-in DeepSeek search provider. It is compatible with DSH 0.1.0-rc.7 and Node.js 20+, allowing configuration of servers, credentials, and result counts through a web settings page. This article covers installation links, credential configuration, provider presets, combination verification, and uninstallation and restoration methods.
Read Moredsh-reef: One-Time Installation for Browser, MCP, and Code Hosting Integration
dsh-reef is a DSH plugin bundle maintained by huey1in, offering five modules with a single installation: native embedded panels, 21 browser_* browser tools, MCP Server, GitHub/GitLab read-only tools, and webhook automatic reviews. It supports multi-tab, multi-profile browser sessions, exposing DSH sessions to external MCP clients, automatic PR/MR reviews, and issue auto-fixing loops. The installation command is dsh plugin --profile web add dsh-reef, licensed under MIT, and requires Node.js >= 22.
Read MoreLocal Shell MCP: Enabling Controlled Shell and File Workspaces for DSH
local-shell-mcp is an MCP control plane maintained by fwerkor, offering clients controlled shell, file, browser automation, and remote worker capabilities. When used as a DeepSeek Harness plugin, it fully bridges the mcp__lsm__ tool interface via HTTP, binding each DSH session to an independent logical session and live workspace. The article introduces the recommended topology, steps to start LSM before installing the plugin, environment variable configuration, and typical usage, and explains that the plugin runs with DSH process permissions, requiring verification of source code and security boundaries before installation.
Read Moredsh-skill-mcp-panel: Unified Management of Skill and MCP in DSH Web Settings
dsh-skill-mcp-panel is a DSH client plugin maintained by Fishquito7, offering two management panels—Skill and MCP—within the web settings page, accompanied by the dsh-panel command line. Skills support workspace partitioning, grouping, batch migration, and scoped enablement/disablement, while MCP enables configuration writes to the cordis.patch.yml managed block via Stdio/HTTP with hot-reloading. It is recommended to install via the v2.0.1 release tarball; after restarting the gateway, the features become accessible on the settings page.
Read MoreArgo: Multilingual Search and Evidence Verification for Agents
Argo (taxuseek/argo) is a multilingual search infrastructure specifically designed for AI Agents, routing queries to 120+ engines by domain and language, and producing concise JSON with confidence scores instead of link lists. It supports one-click installation via DSH, MCP mounting, and command-line invocation, covering vertical scenarios such as finance, film, academia, and code. It is ideal for intelligent agent developers requiring auditable and token-efficient web-connected capabilities.
Read More《Cursor Docs》-MCP Install Links
Looking to share MCP servers, rules, and more? Plugins make it easier to bundle and distribute everything in one package. You can publish to the Cursor Marketplace or your team's private marketplace.
Read MoreThermal Management of MCP and Skills in the Settings Panel via dsh-plugin-manager
Li Qichen's maintained dsh-plugin-manager integrates MCP switch/delete, Skills recycle bin and built-in plugin package browsing into the DeepSeek Harness settings panel, automatically backs up before modifying cordis.patch.yml, and takes effect via host HMR hot update without restarting the dsh web. This article checks the installation commands, capability boundaries and precautions against the community directory page, repository README and source code. The community directory is not an official store, and plugins run with the current dsh process permissions.
Read MoreInstall verified MCP servers for DeepSeek Harness at once with dsh-mcp-bridge
**Translation:** dsh-mcp-bridge is a DeepSeek Harness community plugin maintained by Edge-Echo, which packages the official @deepseek-ai/dsh-mcp-client and a curated set of MCP servers into an installable bundle. By default, only the "everything" demo is enabled, with presets annotated for memory, filesystem, GitHub, Playwright and remote HTTP services. A verify script and CI connectivity checks are added via the servers/ directory definition. The project is categorized under the memory taxonomy. This article covers the directory page, cross-verified installation commands from GitHub and npm, capability boundaries and precautions.
Read MoreManage DeepSeek Harness's MCP servers with dsh-mcp-panel
### 翻译版本1（贴合技术文档正式译法）： DSH-MCP-Panel is a DeepSeek Harness interface enhancement plugin maintained by PerryLink, built upon the official `@deepseek-ai/dsh-mcp-client`: It allows users to check server statuses, tools, errors and reconnection counts via the `/mcp` command and the MCP tab in the settings page, with output desensitized. All add, delete and modify operations are completed via append-only `cordis.patch.yml` snippets, and approvals are written with automatic backups. This article verifies the installation commands, capability boundaries of version 0.4.0 and relevant precautions based on the community directory, GitHub and npm. ### 翻译版本2（更口语化适配社区语境）： DSH-MCP-Panel is a DeepSeek Harness UI enhancement plugin maintained by PerryLink, built on top of the official `@deepseek-ai/dsh-mcp-client`. You can use the `/mcp` command or the MCP tab in the settings page to view each server's status, tools, error logs and reconnection times, with all outputs desensitized. All create, edit and delete operations are done through append-only `cordis.patch.yml` fragments, and approvals will be saved automatically after being written. This article confirms the installation commands, the capability limits of version 0.4.0 and the matters needing attention based on community directories, GitHub and npm.
Read MoreIntegrate OAuth 2.1 MCP into DeepSeek Harness with dsh-oauth-mcp-client
`dsh-oauth-mcp-client` is a DeepSeek Harness community plugin maintained by Springbrand-lab. It adds OAuth 2.1 support (authorization code + PKCE, dynamic client registration, browser login and local callback) based on the native `dsh-mcp-client`. It connects to the Springbrand production MCP Gateway by default, and users can also add other services that support OAuth and Streamable HTTP via the web interface. Tokens are stored by the DSH credential service. It is licensed under MIT, and currently needs to be built locally before being installed into the web profile.
Read MoreManage MCP Servers in DeepSeek Harness Settings Page with dsh-mcp-manager
`dsh-mcp-manager` is a DeepSeek Harness community plugin maintained by hyqhyq3, which manages remote HTTP and local stdio servers in the web interface under **Settings → MCP**. It complements the missing features of the built-in `dsh-mcp-client`, including OAuth (authorization code + PKCE, RFC 7591 dynamic client registration) and stdio transport, with static tokens only referencing environment variable names. Tools are registered in the format `mcp__<name>__*`, optional on-demand broker support is available, and workspace isolation is supported. It is licensed under MIT and requires the web profile.
Read MoreReplace DeepSeek Harness's built-in search with MCP search service using dsh-search-mcp
### 正式翻译版本（符合技术文档语境）： Maintained by gxpppp, dsh-search-mcp replaces the built-in `web_search` backend of DeepSeek Harness with search-class MCPs, while keeping the tool name and display unchanged. This article verifies relevant content through the community directory page and GitHub README, including how to switch between Tavily, Brave, Exa, Perplexity and DuckDuckGo on the web settings page, installation commands, key parsing, uninstallation and restoration methods, as well as the fact that the built-in DeepSeek search will be disabled during activation.
Read MoreAdd DeepSeek V4 protocol adaptation for OpenAI client using deepseek-harness
HenryZ838978-maintained deepseek-harness provides protocol-aware adaptation for DeepSeek V4-Pro / V4-Flash, recording 16 protocol features, 10 contracts and 12 probes. This article verifies the access methods including pip, dsh CLI, MCP and SKILL.md, as well as precautions such as reasoning_content 400, prefix caching and the same-name conflict with the official dsh, based on the community directory and GitHub README.
Read More# agents-sdk Skill: Build Stateful AI Agents on Cloudflare
### Official Translation: This article introduces Cloudflare's official agents-sdk Skill: it guides AI assistants to prioritize retrieving Agents documentation, and correctly build stateful Agents on Workers / Durable Objects, covering features such as `setState`, `@callable RPC`, scheduling, Workflows, MCP and React hooks. The article provides installation steps, agents dependency verification, wrangler configuration and a Counter example, and marks precautions such as decorators, migrations and experimental features.
Read MoreUse notion-knowledge-capture to precipitate AI conversations into a Notion knowledge base
### 译文： Introduce OpenAI's curated Agent Skill "notion-knowledge-capture": This tool relies on Notion's official MCP, and organizes decisions and steps from conversations with tools like Cursor, Codex into linkable Notion pages according to templates such as Wiki, How-To, ADR, and FAQ. The article verifies the installation command, the MCP access methods for Codex/Cursor/Claude Code, as well as the official decision capture and How-To sample workflow.
Read MoreBuild ChatGPT Apps with chatgpt-apps Skill: MCP Server + Widget UI
### 正式翻译结果： ### Introduction to OpenAI's official curated Skill chatgpt-apps: How to scaffold, refactor and troubleshoot ChatGPT Apps SDK applications (MCP server + Widget UI) following a docs-first workflow. The article verifies the capabilities of Skills, application prototype classification, upstream example priority, usage boundaries between MCP Apps bridge and window.openai, as well as the installation and activation methods in Codex and Cursor. It also provides prompt examples for paired use with openai-docs and local joint debugging steps for Developer Mode.
Read MoreTurn Notion research into cited structured documents with notion-research-documentation
### Official English Translation: This article introduces the OpenAI curated skill `notion-research-documentation`: under the premise that Notion MCP has been connected, it enables cross-page retrieval and evidence synthesis within Notion, and generates cited documents in templates such as briefings, summaries, comparisons, and full reports. It follows the official SKILL.md instructions for installation and activation (including Codex's `$skill-installer` and Cursor's skills directory), covers the official five-step workflow, competitive product comparison examples, as well as precautions such as permissions, timeliness, and warehouse deprecation.
Read MoreLinear Skill: Let AI Agents Directly Operate Linear Tickets, Not Just Write Code
### 译文： Officially selected by OpenAI: Skill Linear, which enables Agents to read, create and update Issues, projects and team workflows via Linear MCP. This article introduces its positioning, core capabilities, installation steps for Codex/Cursor, typical usage scenarios such as Bug triage and Standup synchronization, as well as precautions including OAuth and rate limits.
Read More# figma-implement-design: Official OpenAI Skill that 1:1 restores Figma design drafts to production-grade code
**figma-implement-design** is an OpenAI-curated Agent Skill that works with the Figma MCP to translate design nodes into deliverable UI code within repositories. This article introduces its seven-step workflow, boundaries with other Skills such as figma-use, prerequisites for the MCP, as well as installation and typical usage in tools like Cursor (/add-plugin figma) and Codex ($skill-installer), helping developers shorten the Design-to-Code iteration cycle.
Read More# MCP-Builder: Official Anthropic Skill to Walk You Through Building High-Quality MCP Servers
MCP-builder is an Agent Skill in Anthropic's official Skills repository, specifically designed to guide developers in creating high-quality MCP servers. Based on the official SKILL.md and reference documents, this article introduces its four-stage workflow (research and planning, implementation, testing, evaluation), dual-stack support for TypeScript/Python, installation and activation methods (Cursor, Claude Code, etc.), typical prompt examples and usage precautions, to help developers systematically encapsulate external APIs into LLM-callable MCP tools.
Read MoreCursor integrates Gmail/Drive/Calendar: Directly read and write Google Workspace within the IDE Agent
On August 3, 2026, Cursor officially released its Google Workspace plugin, allowing coding Agents to access Gmail, Drive and Calendar within the IDE via Google's remote MCP server. This article sorts out the three major plugin capabilities, OAuth installation path, typical scenarios, as well as permissions and security precautions during the preview period, based on the official Changelog, Marketplace and Google developer documentation.
Read MoreGitHub Trending Highlights: Ponytail and Headroom Lead Agent Context Trimming
On August 4, 2026, the top spots on GitHub Trending were taken by Headroom (a tool that compresses 60-95% of Tokens for output/RAG chunks) and Ponytail (a YAGNI decision ladder that reduces code by 54%). This article verifies the official data of the two projects, sorts out two paths for optimizing Agent context costs: input-side compression (Headroom Library/Proxy/MCP) and output-side restraint (Ponytail Claude Code Skill), and provides access suggestions for combined implementation.
Read MoreThe Two AI Coding Powerhouses in 2026: How Claude Code and Cursor Will Divide Labor
In 2026, a new consensus has formed in the developer community: Claude Code (terminal Agent + Claude Agent SDK) and Cursor (AI IDE) are often used in combination rather than being an either-or choice. Based on official documentation from Anthropic and Cursor, as well as public reviews from Faros.ai, FutureProofing.dev and other sources, this paper compares the differences between the two tools in terms of MCP integration, subagents orchestration, multi-model routing, Bugbot review and token efficiency, proposes a practical division of labor scheme such as "Cursor for daily coding, Claude Code for cross-file refactoring and CI automation", and explains that Claude Code can coexist as an extension of Cursor.
Read MoreMajor Revision to MCP Specification: Protocol Layer Fully Switches to Stateless HTTP
On July 28, 2026, the official MCP team released the largest revision since the protocol was launched: the `initialize` handshake and `Mcp-Session-Id` have been removed, and the core of the protocol has been changed to a stateless request/response model. Remote MCP Servers can be deployed behind load balancers like ordinary HTTP services for horizontal scaling; MRTR now supports elicitation interactions in stateless scenarios; gateway-level routing is implemented via the `Mcp-Method`/`Mcp-Name` headers. This article sorts out the changed mechanism, request comparisons, and key points for developer migration.
Read MoreMajor Version Update of MCP: Stateless Core, Official Extensions and Enterprise-level Certification
On July 28, 2026, the Model Context Protocol released its 2026-07-28 specification, marking the largest revision since the protocol's launch. The core changes include: shifting from a bidirectional stateful protocol to a request/response stateless model, removing the initialize handshake and Mcp-Session-Id; upgrading MCP Apps and Tasks to official extensions; and strengthening OAuth 2.0/OIDC authentication to integrate with enterprise IdPs. With monthly SDK downloads exceeding 400 million and over 950 Claude connectors, MCP is becoming the de facto standard for Agent interconnection.
Read MoreStripe's In-House Minions: End-to-End Coding Agent with Over 1300 Weekly PRs and Zero Manual Coding
In February 2026, Stripe officially disclosed that its self-developed coding agent Minions has been deployed at scale: over 1,300 pull requests (PRs) are fully automated by the agent every week, from Slack commands to passing CI checks, with zero manual code writing and only review required for merging. The core architecture includes an isolated Devbox with 10-second warm-up, a custom Harness forked from Block Goose, hybrid orchestration of Blueprint workflows and agents, the Toolshed centralized MCP tool layer (with nearly 500 tools), and a feedback closed loop of up to two rounds of CI. This article sorts out its design highlights and inspirations for enterprise unattended coding agents.
Read MoreModel Context Protocol Releases July 28, 2026 Update: Remove Session, Move Towards Stateless HTTP Deployment
On July 28, 2026, MCP officially released the 2026-07-28 edition specification. The core of the protocol has been changed from a stateless bidirectional model to a stateless request/response model: the initialize handshake and Mcp-Session-Id have been removed, server/discover and Mcp-Method header routing have been added, and list responses support ttlMs caching. SEP-2567 replaces the Session scope with an explicit state handle, and MRTR retains the mid-interaction capability of tools. Extensions such as MCP Apps and Tasks have been formalized, and AWS Bedrock AgentCore, Netlify and others have expressed their support. This article sorts out the key points of the changes and production migration recommendations.
Read MoreMCP's Largest-Ever Update: Stateless Core Shift on 2026-07-28, Agent Infrastructure Goes Enterprise-Grade
On July 28, 2026, the Model Context Protocol released its 2026-07-28 specification: the core of the protocol has shifted from a bidirectional stateful model to a stateless request/response architecture, the initialize handshake and Mcp-Session-Id have been deprecated, and it now supports any instance handling requests after load balancing. It simultaneously introduces MRTR multi-turn interaction, Mcp-Method/Mcp-Name header routing, and list response cache hints, and graduates MCP Apps, Tasks and enterprise managed authorization to official extensions. OAuth hardening includes RFC 9207 iss verification and DCR migration to CIMD. The monthly downloads of Tier 1 SDKs are nearly 500 million, and companies such as Anthropic and Google Cloud have stated that this is a key step toward the productionization of Agents.
Read MoreMCP Server codebase-memory-mcp: Enabling AI Agents to Truly "Remember" Your Codebase
In July 2026, the GitHub Trending hot project codebase-memory-mcp (produced by DeusData) indexes codebases into persistent knowledge graphs via the MCP protocol. It supports parsing 158 programming languages and provides 15 MCP tools, enabling agents such as Claude Code and Cursor to replace file-by-file exploration with structured queries. According to official benchmark tests, the token consumption for structural queries can be reduced by about 120 times, and indexing for Linux kernel-level repositories can be completed in approximately 3 minutes. This article introduces its Hybrid LSP semantic enhancement, core tools, installation method and applicable scenarios.
Read MoreCopilot Code Review Integrates Agent Skills and MCP: Team Guidelines Finally Can Be Included in Reviews
On July 29, 2026, GitHub announced the general availability of Copilot code review support for Agent Skills and MCP servers, which is now available to Pro, Business and Enterprise users. Teams can inject internal coding and review specifications via the SKILL.md file in the .github/skills directory; MCP can pull external contexts such as Jira and document systems in read-only mode, and attribution tags for skills and MCP are added to review comments. This article sorts out the core capabilities of the general release, the configuration examples of SKILL.md, the read-only security boundary of MCP and implementation suggestions.
Read MorexAI Open-Sources Grok Build: Fully Unveils Harness, TUI and Tooling Layer for Rust Coding Agents
On July 15, 2026, SpaceXAI open-sourced Grok Build (the coding Agent behind the grok CLI and its full-screen TUI) under the Apache 2.0 license. The repository xai-org/grok-build quickly gained over 20,000 stars. Based on official announcements and documentation, this article sorts out the Agent loop, tool layer, TUI, and Skills/MCP extension mechanism, introduces ACP integration, local compilation and custom model configuration, and conducts a horizontal comparison with Codex CLI and OpenCode, to help developers understand the architecture and usage of this heavyweight open-source coding Agent Harness.
Read MoreMCP's Largest Version Update: How Stateless Core Enables AI Agents to Truly Enter Production
On July 28, 2026, the Model Context Protocol released its 2026-07-28 specification, marking the largest revision since the protocol launched: the initialize handshake and Mcp-Session-Id have been removed, and the protocol has shifted to a stateless HTTP request/response model, enabling servers to be deployed on Serverless platforms and behind standard load balancers. MCP Apps and Tasks have graduated to official extensions, OAuth 2.0 authorization synchronization has been strengthened, and full support for Claude will be rolled out gradually. This article sorts out six key points including SEP stateless transformation, extension framework, enterprise authorization essentials, and breaking change migration suggestions for production environment evaluation and reference.
Read MorexAI Open-Sources Grok Build: Full Harness of Terminal-Native Coding Agent Exposed
On July 15, 2026, xAI open-sourced the CLI, full-screen TUI and Rust agent runtime of Grok Build coding Agent under the Apache 2.0 license, and disclosed the complete harness implementations including context assembly, tool calling, Skills/Plugins/MCP extensions and so on. This article sorts out the repository architecture layering, extension system and local construction method, and explains the reasons why it has attracted attention as one of the GitHub AI trending repos in July, as well as how developers can make use of this source code.
Read More