deepseek-aix

English | 中文

deepseek-aix is an AI research assistant for scholarly literature: describe a research topic in conversation, and the agent collects relevant papers through built-in multi-source search (PubMed, arXiv, Semantic Scholar, Crossref, and more), presenting the results as collapsible, bookmarkable paper cards.

Built as a fork of DeepSeek Harness, deepseek-aix keeps the full agent infrastructure — sessions, tools, subagents, workflows, and the Web GUI — and adds a literature capability on top: a multi-source literature-search service, a durable favorites collection with category folders, and a Web UI that renders search results and bookmarks as first-class cards.

Highlights

• Conversational literature search: new sessions start in the deepseek-aix research-assistant mode; describe a topic in natural language (Chinese or English) and the agent searches, deduplicates, and ranks papers.
• Paper cards in chat: every search result renders as a card with authors, year, venue, abstract, and DOI/PMID/arXiv identifiers.
• Favorites with folders: bookmark papers across sessions into a durable, per-user collection, organized into category folders from the sidebar.
• Everything is a plugin: the underlying harness is plugin-based and powered by Cordis, whose design is described in A Programming Paradigm for Spatiotemporal Composability.

Screenshots

Modes & Features

Feature	Literature Search mode	Research mode
Conversational research-topic description (Chinese / English)	✓	✓
Multi-source literature search (PubMed, Europe PMC, bioRxiv/medRxiv, Crossref, OpenAlex, Semantic Scholar, arXiv)	✓	—
Automatic deduplication and cross-round merging	✓	—
Paper cards (authors, year, venue, abstract, DOI/PMID/arXiv, open-access badge)	✓	—
Favorites with category folders (durable across sessions)	✓	—
Full-text retrieval (open-access subset)	✓	—
Literature-search methodology skill	✓	—
Workspace and file management	✓	✓
Sandboxed code execution (bash / PowerShell)	✓	✓
Subagents and parallel research	✓	✓
Goal auto-continuation	✓	✓
Plan mode and task breakdown	✓	✓
Durable sessions, replay, and multi-model configuration	✓	✓

Developer preview

deepseek-aix is currently in developer preview and is iterating rapidly. THERE WILL BE COMPATIBILITY-BREAKING CHANGES.

Run

Run from npm

Install Node.js, then run:

npx @deepseek-ai/dsh web --port 3090

The command starts the Web UI at http://127.0.0.1:3090 (the port is configurable; the default is 3080). See Web UI guide.

Run from source

To run from a repository checkout:

git clone https://github.com/deepseek-ai/deepseek-harness.git
cd deepseek-harness
pnpm install
pnpm run build
pnpm dsh web

Community and support

• Feel free to submit feedback or bug reports through GitHub Discussions.
• Join DeepSeek Harness Discord community.

Contributing

See CONTRIBUTING.md.

Development

Start with the development guide and architecture documentation.

For agents, follow AGENTS.md.

License

MIT

Third-party dependencies and their licenses are disclosed in THIRD_PARTY_NOTICES.md.