Let your AI agent map a research field: the papers it is built on and the new ones

An MCP server: works with Claude Code, Cursor, and Codex.

Give your agent a topic or a few papers, and it gets the papers around it and which ones cite which. From that you see which papers the field is built on and which recent ones build on them. The data comes from OpenAlex and Semantic Scholar.

Try it

ASK YOUR AGENT

What are the foundational papers on speculative decoding, and which recent papers build on them?

Behind the scenes

omniseek_field_skeleton(query="speculative decoding", source="s2")

WHAT CAME BACK

2022, cited by 17 in the mapFast Inference from Transformers via Speculative Decoding
2023Accelerating Large Language Model Decoding with Speculative Sampling
2025, cited by 15 in the mapReward-Guided Speculative Decoding for Efficient LLM Reasoning

From a real run on 2026-10-10 (60 papers, 198 citation links). Each paper comes back with its year and how many papers in the map cite it; your agent sorts them into foundations and recent work.

It uses OpenAlex by default, and Semantic Scholar when you want the newest arXiv papers. Both have free daily limits; when one is busy the tool says so, and your agent can switch to the other.

Install

TWO COMMANDS

uv tool install omniseek
claude mcp add omniseek -- omniseek

Python 3.11 or newer; uv installs it if you don't have it. Using Cursor, Codex, or another setup? See Install.

Where it looks

openalex: the open scholarly graph
semantic_scholar: papers, citations, and citing sentences
arxiv: preprints
crossref: published works by DOI

Every source, with whether it needs a key or a login: docs/sources.md. Tool reference: docs/tools.md.

OmniSeek · GitHub · PyPI · Apache-2.0