The "coding-agent" topic on GitHub groups 148 open-source projects in the DeepSeek Harness (DSH) ecosystem, led by planning-with-files with 26.8k GitHub stars. planning-with-files — Persistent file-based planning for AI coding agents and long-running tasks. Every project here is indexed by DSH Universe with live GitHub data — stars, activity and install status — so you can compare and install directly.
Persistent file-based planning for AI coding agents and long-running tasks. Crash-proof markdown plans, session recovery after /clear and compaction, per-turn re-injection against context rot, deterministic completion gate. Manus-style. Install from npm, the Claude Code plugin marketplace, or npx skills. Codex, Cursor, OpenCode, 60+ agents.
✨ All your agents and workspaces in one place, on every device you own. Track tasks on a board, accessible from desktop, mobile, browser, or API. Self-hosted.
TUI companion plugin featured in the DSH official WeChat account: Claude Code style, whale top bar / live status / streaming thoughts / double-Esc rollback / context progress + TPS. One-click npm install. DSH official WeChat featured TUI plugin — Claude Code style: whale bar, live status, streaming thoughts, double-Esc rollback, context bar + TPS. npm one-click.
HarnessRouter Community Edition: the self-hosted, Apache-2.0 edition of the unified interface for agent harnesses. Run Codex, Claude Code, Hermes, PI, DSH, and more through one API, with sessions, streaming, files, cancellation, and failure handling. Implements the Unified Harness Protocol (UHP), an open standard. Your keys, your infrastructure.
No ROOT, no Termux: run DeepSeek Harness on your phone. Full Ubuntu environment + proroot with zero ptrace overhead · AI output on screen in real time · direct ADB connection · no data loss
A curated collection of skills for AI coding agents. Skills are packaged instructions and scripts that extend agent capabilities across development, documentation, planning, and professional workflows.
Agent Skills that teach coding agents to write, review, and port idiomatic Rust, plus a tested clippy config. Works with Claude Code, Codex, Copilot, Cursor, and Gemini CLI.
Tianshu is a terminal coding agent runtime built on the harness project (Tui X Gui), with prefix-cache engineering optimization for DeepSeek V4 (measured steady-state hit rate of 97–99% in long sessions) and deep adaptation. It breaks out of the traditional AI coding assistant's limit of treating the LLM merely as a tool, and is built on a cognitive virtual machine (CVM), a self-awareness layer and Stigmergy self-decaying memory, making the AI a development partner with independent judgment and cognitive safeguards.
The design skill for Claude Code, Cursor and any coding agent. Stop shipping AI-slop UI: turn it into shippable, tasteful frontend. Install: npx skills add superdesigndev/superdesign-skill. Powered by superdesign.dev
DeepSeek Harness (DSH) plugin: a live iOS Simulator — and a USB-connected iPhone — inside the conversation. 22 agent tools for booting, building, driving the UI by accessibility identity, OCR text or list rows, plus a streaming sidebar panel you can tap and drag on.
A curated list of plugins, skills, MCP servers, patch/profile layers, orchestrators & UIs for DeepSeek Harness (DSH). Visualization · PPT · Coding · Agents · Loops (auto-research) and more. #dsh
DeepSeek Harness plugin for Android — build, run, and interact with a live emulator or USB device stream inside a conversation, driven entirely through adb.
A proactive multi-session coding-agent harness on herdr + omo: one lead, a duty loop, many lanes, a gated merge. Plus Dori mode, the messenger-facing skill.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
Claude Code-style auto-memory plugin for DeepSeek Harness (dsh): typed memory files + MEMORY.md index auto-injected into the system prompt. File-only, no external services.
Deploy and host DeepSeek Harness (dsh) instances — self-hosted on your own server or cloud-hosted: a multi-tenant platform with isolated Docker containers, persistent workspaces and resource quotas.
Oh My Tianshu is built on dsh harness with a TDD workflow, cross-session memory, verification gates, agent routing, semantic + graph code retrieval, file rollback and a full-screen terminal UI — all assembled as plugins. It is a friendly MIT fork of DeepSeek Harness, keeping the upstream everything-is-a-plugin architecture, and an independent release that reworks and merges the core components of Tianshu harness.
Mobile client for the DeepSeek Harness — use the dsh on your dev machine from your Android or iOS phone, remotely across networks, peer-to-peer over iroh with no server to set up (falls back to a relay that only sees ciphertext)
dsh-filesnap — rewind a conversation and the files it changed together to before a given turn, no git repository needed. A blazing-fast rewind and redo plugin for DeepSeek Harness, powered by a 🦀 Rust core, tracking the conversion and the files it changed, no git required, low disk consumption
The Feishu UI for DeepSeek Harness — a panel-driven control console: every slash command a button on the ⚙️ control-panel card, in-card approvals & questions, live streaming cards, one-QR setup. | Feishu UI for DeepSeek Harness: a panel-driven console — every command is a card button, in-card approvals & questions, streaming cards, one-QR setup.
oh-my-dsh aims to be the best TUI companion for DeepSeek Harness. It brings Harness to the terminal with familiar behavior and interactions, careful visual design, and a focus on making everyday coding feel clear and fluid. Built on the published Harness plugin runtime, it runs as omdsh.
Minimal surface, complete architecture: a small, local-first coding-agent harness that re-derives the invariants of DeepSeek Harness with fewer concepts. Independent educational project, not affiliated with DeepSeek.
DeepSeek Harness Android app: control your DeepSeek Harness agents from your phone. The agent runs on your PC; a host plugin inside DSH pairs the phone by QR code over Wi-Fi or Tailscale, no cloud. Chats of all projects, live replies, photos and files, steer, agent questions, notifications. Unofficial, MIT.
Personal Agent Skills and DeepSeek Harness (DSH) plugins: message edit, Skill Scout, prompt recall, autosuggest and enhance, plus a DSH plugin index. /init for Pi, DSH, Cline, Grok Build, goose and Aider, with presets from Claude Code, Codex, OpenCode, Crush, Qwen Code, Kilo Code, Kimi Code and Copilot CLI.
🕊️ Run a full DeepSeek Harness coding agent inside Feishu/Lark — native thinking process (CoT), interactive approval cards, live reactions, slash commands, WS long-connection. No public callback URL.
DeepSeek Harness plugin that does QA: an agent explores your app like a real user and leaves evidence, then the explored path is exported as a deterministic replay scenario you run every release. Orchestrates the browser, macOS desktop, iOS and Android drivers — never a false green.
A portable DeepSeek Harness setup: sixteen plugins that make an agent work in small verified steps (second-model review, loop recovery, lazy MCP, local memory), plus a one-command installer.
A desktop agent app built on the DeepSeek Harness (dsh) runtime — an Electron shell bundles the dsh CLI, provisions built-in plugins and skills, and hosts the dsh Web UI. Build your own desktop agent app through plugins.
Project-level change routing and reuse governance for AI coding agents. Helps Codex and Claude Code find the right capability boundary, avoid duplicate implementations, respect write constraints, and calibrate routing decisions for large full-stack repositories.
DeepSeek Harness (DSH) plugins. qwen-image gives a text-only coding model eyes: an image goes to a Qwen-VL route through ctx.llm and comes back as text, so DeepSeek keeps coding while Qwen looks. Pure ESM, no build permission at install. | DSH plugin set: qwen-image lets text-only models read images via Qwen VL and return text; pure ESM, no build authorization needed at install.
AgentCabin — open-source desktop workspace for Pi Coding Agent and DeepSeek Harness (DSH), built with Electron, Svelte, and Rust. Local-first AI coding agent desktop for Pi Coding Agent and DeepSeek Harness (DSH), built with Electron, Svelte, and Rust.
Context housekeeper for DeepSeek Harness: remembers “what you're doing right now” for you, automatically curating the context sent to the model — keep what should be kept, delete what should be deleted, leave uncertain items untouched, so long AI coding sessions stay on track and every token counts.
A full engineering bench for your coding agent: the work survives any session, and every done comes with proof. Runs on Claude Code, Codex CLI, Copilot CLI, Kiro CLI, and DeepSeek Harness.
A DeepSeek Harness bundle reimplementing opengsd-core (Git Ship Done) as host-plane Cordis plugins, replacing the default agent-loop behaviour with the GSD phase loop.
Le dsh-proof est un plugin qui vérifie automatiquement le code après chaque modification (types, qualité et tests) et oblige l’agent IA à corriger les erreurs avant de terminer son travail.
DeepSeek Harness plugin: role-based model routing — planner (root agent) on deepseek-v4-pro, delegated executor subagents on deepseek-v4-flash; ships a prompt section and a pro-flash-routing skill.
Project documentation system skill centered on improving what exists: optimize existing project docs, solidify AI role-based collaboration workflow; also supports joining mid-way to build up the system and new project initialization. Question-driven, 26 templates, MIT open source.
DeepSeek Harness (dsh) on your computer, DeepSeek-V4.1-Flash 1017 or Kimi K3.1 pro on the GPUs of the ROMEO supercomputer: one command starts the model, the tunnel and dsh.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
ARI (Agent Runtime Interface): one shell, any coding-agent harness. An interface specification derived from source-level study of DSH, Codex, ZCode, OpenCode, Pi, and ACP.
🌉 Move coding-agent sessions between harnesses — DeepSeek Harness (dsh), Codex Desktop, Claude Code, ZCode, agy. Lossless tool calls, reasoning and timestamps through one normalized model. Written in Rust.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
Every file, one version before it breaks. A DSH plugin: snapshots a file the instant before an agent rewrites it, with a per-session GUI timeline where any point can be restored in one click.
DeepSeek Harness plugin: pick the model your subagents run on — with a reasoning-effort ceiling and an output-token cap — enforced in code on every child DSH starts, whichever tool starts it.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
A preventive governance plugin for DeepSeek Harness: the cloud planner is denied write and shell access, a local worker implements inside a bounded workspace, and promotion is hash-bound against a machine-checked or operator-attested record. Five boundaries, each failing closed.
Run live iOS simulators and your real iPhone over USB inside DeepSeek Harness conversations with 22 agent tools, MJPEG previews, and SwiftUI hot reload.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
Plugin bundle and launcher for DeepSeek Harness (dsh): lower cost per coding task, verify-after-edit, owned paths, model flags. Benchmarked against stock DSH, Steak Pi and Claude Code.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
RTK command rewriting and tool-output compaction for the DeepSeek Harness: rewrites bash commands to their rtk equivalents and compacts the result before the model sees it.
Before you close an AI coding session: one verdict on whether to archive or delete it - dirty repos, unpushed commits, scratch files, and whether the session needs a handoff.
DeepSeek Harness (DSH) plugin: dispatch work to DSH agents from Claude Code / Codex — native subagent progress, in-host worker sessions with per-tier presets, and a multimodal bridge that lends the text-only harness vision and image generation.
DeepSeek Harness plugin for macOS desktop automation. Observes the accessibility tree, gates risky actions behind one-time approval, permanently refuses secure fields, and returns bounded visual evidence — through a native helper you build and grant locally.