Sciverse Agent Tools MCP
An MCP server and skill for Sciverse paper search: metadata, semantic retrieval, citations and full text for agents.
- 129 ★ on the repository
- Last commit
Extended access
Why this level
- Queries and the token go to the external Sciverse Open Platform service.
- The server only reads data and changes nothing in your files or accounts.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/sciverse-agent-toolsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Queries and the token go to the external Sciverse Open Platform service.
- The server only reads data and changes nothing in your files or accounts.
This is third-party code. Review the repository files before installing.
What it does
Sciverse Agent Tools connects an agent to the Sciverse Open Platform academic database. Six tools split the work: `list_catalog` shows available fields and filter operators, `search_metadata` searches papers, authors, sources and patents, and `semantic_search` finds text passages from a natural-language query. `list_paper_relations` pages through a paper's citations and references, `read_content` returns a slice of the source document, and `get_resource` fetches figures and tables referenced in the Markdown. The same set is available through an MCP server, a Claude Code skill, a ClawHub skill and Python and TypeScript SDKs.
Who it is for. Researchers and developers building literature reviews and scientific RAG with citations to original sources.
Good fit when
- An agent assembles a literature review and searches papers with structured filters.
- An answer needs exact passages from papers for citations and verification.
- You need to walk through the citations and related works of one paper.
- An agent pulls figures and tables from a paper for a multimodal answer.
Not a fit when
- Queries cannot leave your machine: every call goes to the Sciverse Open Platform.
- You have no Sciverse token, and all six tools require authentication.
- You need Russian-language sources: Sciverse describes a general academic base with no language promises.
Example request
Use Sciverse to find papers on attention mechanisms in transformers, show passages and open the full text of the most relevant one.MCP
- Transport
- stdio
- Authentication
- API key
Environment variables
SCIVERSE_API_TOKENrequired, secret- Bearer token from the Sciverse developer console
Limitations
A `sv-...` token from the Sciverse developer console is required, and the platform is external: queries and texts pass through its service. The server needs Node.js with `npx`, and the SDKs need Python or Node.js. GitHub metadata lists the repository license as undetermined (NOASSERTION), so check the LICENSE file before using it in a project. Offsets in `read_content` count Unicode code points, not bytes. Figures and tables from `get_resource` arrive as image bytes and take noticeable agent context.
How to disable. Run `claude /plugin uninstall sciverse` or remove the `sciverse` entry from `.mcp.json`. Delete the `~/.sciverse/credentials.json` token file separately.
FAQ
How do I connect Sciverse to Claude Code?
Run `claude /plugin marketplace add https://github.com/opendatalab/Sciverse-Agent-Tools` and `claude /plugin install sciverse`. The plugin installs the skill and the MCP server, and reads the token from `SCIVERSE_API_TOKEN` or `~/.sciverse/credentials.json`.
How do I connect the server to Cursor or Codex?
Add a launch of `npx -y sciverse-mcp-server` with the `SCIVERSE_API_TOKEN` variable to Cursor's `.mcp.json` or to `~/.codex/config.toml`. Step-by-step guides live in `docs/integrations/`.
What does semantic_search do?
It finds passages of paper text from a natural-language query for RAG. To widen context around a hit, call `read_content` with offsets.
Where do I get a token?
In the Sciverse developer console on sciverse.space. Set it with `SCIVERSE_API_TOKEN` or the `sciverse auth login` command from the Python SDK.
Which tools does the set include?
Six: `list_catalog`, `search_metadata`, `semantic_search`, `list_paper_relations`, `read_content` and `get_resource`. Start with `list_catalog` to learn the available fields and filter operators.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server for searching and downloading academic papers from arXiv, PubMed, bioRxiv, Semantic Scholar and other open sources
PubMed MCP searches biomedical papers, retrieves available full text and formats citations with retraction and correction notices.
A Claude Code plugin and MCP server: over 1000 scientific tools for biology, chemistry and medicine, data-check commands and routing skills
The MCP server, CLI and SDKs of the QVeris platform: the agent finds a data service, calls it and audits credit spending
Connects a Zotero library to Claude and other agents: search papers, annotate, cite, and search by meaning
Official MCP server from the OpenSearch project exposes a cluster to agents: indexes, mappings, DSL search and shard state.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Sciverse Agent Tools (this card) | no | API key | 129 | |
| Paper Search MCP | no | key is optional | 3,008 | searches several open sources; Sciverse works with one platform and adds passage-level semantic search and full-text slices |
| arxiv-mcp-server | no | not required | 3,258 | reads arXiv papers; Sciverse covers papers, authors, sources and patents under one authentication |
README in short
All six tools share Bearer-token authentication and one `openapi.yaml` schema. For code there are `sciverse` SDKs on pip and npm, a GPT Researcher plugin and cookbook notebooks for literature reviews, deduplication and citation checks.