#7272nd of 89 in Agent platforms
dsha
Android app that runs DeepSeek Harness in a bundled Ubuntu rootfs, no root
Documentation ↗
(opens in a new tab)Website ↗
(opens in a new tab)Repository on GitHub ↗
(opens in a new tab)
- Stars
- 3.2k
- License
- MIT
- Last commit
- Nov 2026
- Last release
- Oct 2026
- Language
- Java
Missed the latest GitHub check; these facts may be older.
Overview
DSHA is an Android APK (Java, arm64-v8a, Android 11+) that bundles an offline Ubuntu 24.04 rootfs with Node.js 24 and DeepSeek Harness 0.2.2-rc.1, run through proot or optionally proroot. It serves the harness Web UI and adds a local bridge on 127.0.0.1:3090 so the agent can read the screen, tap and run device commands via accessibility, ADB, Shizuku or root. It also ships a PTY terminal, plugin manager, backups and a LAN proxy.
Who it is for: Android users who want DeepSeek Harness on a phone with device control
Strengths
- Offline Ubuntu rootfs in the APK: apt, PTY and native modules work without Termux
- Accessibility-based screen control needs no ADB, Shizuku or root
- Built-in wireless ADB pairing; one command whitelist shared by ADB, Shizuku and root
- Scheduled backups, startup diagnostics and a separate emergency DSH environment
Weaknesses
- APK is about 302 MiB because the full Ubuntu environment is bundled
- arm64-v8a only; Android 10 and below are unsupported
- No bash sandbox isolation; default permission mode is danger-full-access
- Virtual screen and privileged execution are experimental and device-dependent
What it needs
- no GPU
- Needs deepseek-harness (@deepseek-ai/dsh), proot / proroot, Shizuku or Stellar (optional)
- Models: Provider, model and API key configured in the harness Web UI
Also in Agent platforms
See all 89| Rank | Project | Score |
|---|---|---|
| 1 | AutoGPTRole-based AI agents coordinated by one lead agent, plus a visual workflow builder | 95 out of 100 |
| 2 | daily_stock_analysisLLM-driven daily analysis of watchlist stocks across six markets, pushed to chat apps | 95 out of 100 |
| 3 | LangflowVisual builder for AI agents and workflows, exposed as API or MCP server | 94 out of 100 |
| #4 | deer-flowAgent harness running sub-agents, memory and sandboxes with pluggable skills | 93 out of 100 |
| #5 | DifyVisual builder for LLM workflows, RAG pipelines and agents, with APIs | 92 out of 100 |