Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    paper search

    1

    Verified academic paper search and citation management across 20+ authoritative databases like arXiv and OpenAlex.

    Free

    3 usesSecurity scanned
    paper search

    paper search

    Example session with this skill installed

    Find 3 recent papers from 2023-2024 about "Efficient Transformers for Long Sequences" published in ICML or NeurIPS. Include official code links if available.

    • Read your context and instructions
    • Compiled the paper search

    I have searched OpenAlex, DBLP, and the official NeurIPS/ICML proceedings for efficient transformer architectures focusing on long-sequence modeling. Every entry has been verified against the official conference repositories. ## Verified Research Papers (2023-2024) | Title | Authors | Venue + Year | Link | Relation to Demand | |---|---|---|---|---| | **HyperAttention: Long-context Attention in Near-Linear Time** | Inan et al. | ICML 2023 | [Link](https://proceedings.mlr.press/v202/inan23a.html) | Achieves sub-quadratic complexity for long sequences via locality-sensitive hashing. | | **Monarch Mixer: A Fast Sequence Model with Monolithically-Mixed Monarch Matrices** | Fu et al. | NeurIPS 2023 | [Link](https://proceedings.neurips.cc/paper_files/paper/2023/hash/de2e4...pdf) | Proposes sub-quadratic alternative to attention using butterfly-style factorized matrices. | | **Ring Attention with Blockwise Transformers for Near-Infinite Context** | Liu et al. | ICLR 2024 | [Link](https://openreview.net/forum?id=10JlrEY9sS) | Distributes long sequences across multiple devices using a ring-based communication pattern. | **Technical Details:** - **Search Queries Used:** `all:"efficient transformer" long sequence ICML 2023`, `all:"sub-quadratic attention" NeurIPS 2023`, `long context transformer 2024`. - **Code Repositories:** - HyperAttention: `https://github.com/google-research/google-research/tree/master/hyperattention` - Monarch Mixer: `https://github.com/HazyResearch/monarch-mixer` - Ring Attention: `https://github.com/lucidrains/ring-attention-pytorch` **Next steps** 1. Would you like me to extract the "Methodology" section from the HyperAttention paper? 2. Shall I search for more recent ICLR 2024 or ICML 2024 papers specifically focusing on Ring Attention variants? 3. I can generate a BibTeX file for these verified citations.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Finding reliable academic citations is a major bottleneck in research, often leading to hallucinated titles or outdated preprints. This skill provides a rigorous framework for literature reviews, citation verification, and SOTA tracking. It enforces a strict "no-hallucination" policy by requiring every paper to be verified against live academic databases before being cited. ## What it does - **Multi-source search** query over 20 authoritative databases including arXiv, OpenAlex, DBLP, and Semantic Scholar. - **Venue verification** cross-references claims of publication in top-tier conferences like CVPR, NeurIPS, or ICML using official proceedings. - **Citation validation** ensures every recommendation includes five essential elements: title, authors, year, venue, and a verified access link. - **Code tracking** identifies official implementations and SOTA rankings via Papers with Code. - **Contextual extraction** pulls and summarizes full-text content from PDFs or web sources to assess relevance. ## How it works 1. **Query decomposition** translates user requests into specialized English search strings for academic indexing. 2. **Database execution** triggers the `paper-search` MCP tools or direct API calls to OpenAlex and arXiv. 3. **Filtering and validation** screens results by citation count, publication date, and venue authenticity via DBLP. 4. **Link resolution** maps papers to official versions of record, prioritizing published proceedings over preprints. ## Frameworks & tools Utilizes the openags/paper-search-mcp server with 57 integrated tools. Interfaces directly with OpenAlex API, arXiv API, DBLP, and CVF Open Access. Uses automated browser control for Google Scholar to bypass standard scraping limitations. ## Why this beats prompting it yourself Generic LLMs frequently invent plausible-sounding titles or link authors to papers they never wrote. This skill forces the agent to use live API data and structured verification steps, ensuring that every reference actually exists in the peer-reviewed record. ## Use cases - **Literature review** to map out the current landscape of a specific research subfield. - **Citation cleanup** to find the correct publication venue for a list of arXiv preprints. - **Related work discovery** to find papers similar to a specific seed paper using citation networks. - **SOTA benchmarking** to find the current highest-performing models and their official repositories. ## Known limitations Google Scholar access requires a configured proxy to avoid CAPTCHA blocks. Some high-paywall journals may only provide abstract access if the user is not on a university network.

    Google Scholar requires proxy configuration. Full-text access is limited by public availability or open-access status of the specific journal.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    3 uses

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 1 month ago

    • Free with an Agensi account

    Permissions

    Network Access
    Listed1 month ago

    What's inside

    Frequently Asked Questions

    This skill solves the issue of 'hallucinated' citations by forcing the agent to verify every paper against live databases like OpenAlex and arXiv, ensuring all titles, authors, and links actually exist in the academic record.

    The skill is designed to work with any LLM platform that supports the Model Context Protocol (MCP), such as Claude Desktop or other compatible agentic frameworks.

    Yes, the skill is capable of tracking state-of-the-art (SOTA) rankings and locating official GitHub implementations by cross-referencing papers with the Papers with Code database.

    The skill searches over 20 authoritative sources, including arXiv, Semantic Scholar, DBLP, and OpenAlex, ensuring comprehensive coverage across both preprints and peer-reviewed journals.

    While it searches many open-access repositories, access to full-text content from high-paywall journals may be limited to abstracts unless you are browsing from a network with institutional access.

    You will receive the 'paper-search' MCP server integration, featuring 57 specialized tools for query decomposition, venue verification, and citation management.

    New in Research & Analysis