MCP server
deepseek-ai / deepseek-harness
DeepSeek Harness: Everything is a Plugin.
This connects Repowise to deepseek-ai/deepseek-harness's code. It does not install deepseek-harness.
Repowise maps public repos from their source. Not affiliated with deepseek-ai. What is Repowise?
What is deepseek-harness?
deepseek-harness is an open-source agent harness developed by DeepSeek AI for running DeepSeek-backed agents locally. It is built on an everything-is-a-plugin architecture powered by Cordis. Given a user's prompt and workspace, it runs a session-based agent loop that calls DeepSeek models and sandboxed tools, and it produces persisted conversations, file deliverables and session logs. The project is written mainly in TypeScript. To install it, run git clone https://github.com/deepseek-ai/deepseek-harness.git and follow the steps below. Its code health is 7.6 out of 10 (Good), ranked #96 of 4,000+ indexed repos. The last commit to deepseek-harness was on Oct 23, 2026, and 62 people have contributed.
As of commit 22ddee3, indexed Oct 24, 2026. What it is comes from a model reading the README; the rest is measured from the source and git history, not from the maintainer.
For agentsdeepseek-harness.md
Repo facts: type, license, health, rank, contributors
| Type | MCP server |
|---|---|
| License | MIT |
| Code health | 7.6 / 10 (Good) |
| Health rank | #96 of 4,000+ indexed repos |
| Last commit | Oct 23, 2026 |
| Contributors | 62 |
| Bus factor | 3 |
| Size | 17,318 files, 1,106,519 lines |
Install deepseek-harness
Commands copied from the README
- 1Clone the deepseek-harness repository.
git clone https://github.com/deepseek-ai/deepseek-harness.git - 2Move into the deepseek-harness folder.
cd deepseek-harness - 3Install Node.js, then start the Web UI with npx.
npx @deepseek-ai/dsh web - 4Install the repository dependencies with pnpm.
pnpm install - 5Build the repository artifacts, then start the Web UI from the built artifacts.
pnpm run build
Explore deepseek-harness
- DocumentationModule-by-module documentation, 5,212 pages.
- ArchitectureDependency graph, layers and third-party dependencies.
- Code healthEvery file's score and the findings behind it.
- CommitsEach commit's effect on code health.
- OverviewEverything else the index found: contributors, decisions, what changed this week.
Ask about this code in your agent
This connects Repowise to deepseek-harness's code. It does not install deepseek-harness. To install deepseek-harness, use the install steps.
Useful if you want to change or fork deepseek-harness: your agent can ask how it is put together and get answers that cite its files.
Ask about deepseek-harness
Answers come from deepseek-harness's README and source, with the files they cite. Two free answers a day without an account.
Want this for your own codebase? Map your repo free
About this page
Get a page like this for your own codebase
Repowise built this map of deepseek-ai/deepseek-harness from its source and git history. It parses every symbol, draws the dependency graph, scores each file's health from complexity, duplication, coverage and churn, finds hotspots and owners in the history, and writes documentation for every module. Your coding agent reads the same index over MCP, so its answers cite real files.
Free for 2 public repos, with a page like this one. No card. See how Repowise works
Prefer local? pip install repowise && repowise init indexes your own checkout, no account needed. Star Repowise on GitHub
deepseek-ai/deepseek-harness FAQ
What is deepseek-harness?
deepseek-harness is an open-source agent harness developed by DeepSeek AI for running DeepSeek-backed agents locally. It takes a user's prompt and workspace and runs them through a session-based agent loop.
How does deepseek-harness work?
deepseek-harness runs a session-based agent loop that calls DeepSeek models and sandboxed tools. It is built on an everything-is-a-plugin architecture powered by Cordis and produces persisted conversations, file deliverables and session logs.
How do I install deepseek-harness?
deepseek-harness can be started with npx @deepseek-ai/dsh web after installing Node.js. It can also be run from a repository checkout with pnpm install, pnpm run build and pnpm dsh web. See the install steps
What language is deepseek-harness written in?
deepseek-harness is written mainly in TypeScript. The repository also contains JavaScript, Python, Swift, C++ and shell files.
How is the deepseek-harness code organized?
deepseek-harness is organized into modules such as apps, packages, scripts and website. The apps directory holds the CLI, desktop and plugin market generator entry points, and packages holds the core modules and extensions.
Who is deepseek-harness for?
deepseek-harness is aimed at developers who want to run DeepSeek-backed agents locally and build on them. It is in developer preview, and compatibility-breaking changes are expected.
What license does deepseek-harness use?
deepseek-harness uses the MIT license, identified by the SPDX id MIT.
Code health
Hotspots average 6.7, the place fixes pay off most.
On the leaderboards:#3 of 6 Coding agents#37 of 70 TypeScript repos
Key features of deepseek-harness
- Built on an everything-is-a-plugin architecture powered by Cordis.
- Starts a Web UI with the npx @deepseek-ai/dsh web command and opens it in the default browser for local launches.
- Accepts the --no-open option to run the server without opening a browser.
- Can be run from a repository checkout by installing dependencies, building the artifacts and running pnpm dsh web.
- Supports plugin authors, who can add the dsh-plugin topic to their plugin repositories for discoverability.
How deepseek-ai/deepseek-harness works
Input enters through one of the front ends: the CLI entry point at apps/cli/src/bin.ts, the desktop app started from apps/desktop/src/main.ts, or the Web UI started with the dsh web command. The core work happens in the agent loop defined in packages/core/agent-loop/src/agent.ts, which calls DeepSeek models and sandboxed tools for each session. The packages directory is the largest part of the repository, with 5107 files, and holds the core modules and extensions that the loop and the front ends build on. Because everything is a plugin, features are added as plugins built on Cordis. The scripts directory holds the scripts that generate, check and verify the documentation and packaged output, reading the source tree and writing catalog pages or reporting violations. Each session ends with persisted conversations, file deliverables and session logs.
The codebase has 17,318 files and 48,698 symbols in 11 modules, led by TypeScript, JavaScript and Python. Code health is 7.6 out of 10, rated good.
Maintain deepseek-ai/deepseek-harness?
Claim this page, add a badge to the README, or re-index it.
Claim this page
Sign in with GitHub and we check that you are an admin or maintainer of deepseek-ai/deepseek-harness. Then add a badge to the README that links readers here.
Add a badge to your README
Readers click through to this page and can ask deepseek-harness's code a question. A badge also re-indexes the repo every week and marks you as the verified maintainer.
[](https://repowise.dev/repo/deepseek-ai/deepseek-harness?src=badge_ask)Paste it at the top of README.md, under the title.
Added it already? Verify to get the mark now instead of tomorrow.
Need HTML or the larger health card? More formatsHide more formats
This badge as HTML
<a href="https://repowise.dev/repo/deepseek-ai/deepseek-harness?src=badge_ask"><img src="https://api.repowise.dev/badge/wiki/deepseek-ai/deepseek-harness.svg?label=ask" alt="repowise: Ask this codebase"></a>Code health card
The score, its change since the last index, and when it was measured. Follows the reader's light or dark theme.
<a href="https://repowise.dev/repo/deepseek-ai/deepseek-harness?src=badge_card"><picture><source media="(prefers-color-scheme: light)" srcset="https://api.repowise.dev/badge/health/deepseek-ai/deepseek-harness.svg?style=card&theme=light"><img src="https://api.repowise.dev/badge/health/deepseek-ai/deepseek-harness.svg?style=card" alt="repowise code health" width="500" height="168"></picture></a>