Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Scientific Research Copilot

    1

    Turn a general coding agent into a lab-ready research assistant that picks the right scientific tools and produces reproducible, reviewable results.

    Free

    6 usesSecurity scanned
    Scientific Research Copilot

    Scientific Research Copilot

    Example session with this skill installed

    I have a 10X single-cell dataset from tumor biopsies in data/filtered_matrices/ (matrix.mtx, barcodes.tsv, features.tsv for 3 samples). Using the available research skills, run QC, remove doublets, normalize, cluster, annotate major populations with marker genes, and produce a labeled UMAP plus a cluster marker table. Save the processed object and intermediate files under results/, pin all library versions and random seeds, and finish with a short methods note I can drop into my manuscript draft. Flag any batch effects between samples before I trust the clustering.

    • Read your context and instructions
    • Compiled the scientific research copilot

    Summary

    Completed an EGFR inhibitor discovery pass: retrieved 148 known actives from a public chemical database, filtered to 41 drug-like candidates by molecular weight and logP, and ranked the top 6 by scaffold similarity and docking readiness. Every step is logged with library versions and query dates.

    CandidateMW (g/mol)logPScaffold sim.Docking-ready
    quinazoline A383.42.70.88yes
    pyrimidine B411.53.40.81needs salt form
    anilide C442.64.20.77yes
    extended scaffold D498.75.10.69flagged: MW outlier

    Environment: Python 3.11, rdkit 2024.3, fixed seed 42; database snapshot retrieved 2026-09-12, responses cached in data/cache/.

    Next steps

    • Inspect results/candidates.svg for the scaffold clustering plot (vector format, colorblind-safe palette)
    • Verify the two lowest-similarity hits against a second source before ordering compounds
    • Prepare protonation states for pyrimidine B, then export the docking batch from prep/

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

    About this skill

    The problem

    General-purpose agents often hallucinate scientific API endpoints, fail to handle complex bioinformatics file formats, or ignore reproducibility standards. This leads to broken pipelines and untrustworthy results in data-heavy research workflows.

    What it does

    • Provides curated entry points for genomics, cheminformatics, and clinical data APIs.
    • Automates structured processing for FASTA, VCF, BAM, and molecular structure files.
    • Generates publication-ready figures with proper units, legends, and colorblind-safe palettes.
    • Enforces reproducibility by logging library versions, seeds, and intermediate data states.

    Frameworks & tools

    Python 3.11+, RDKit, Biopython, Scanpy, Pandas, Matplotlib, Scikit-learn, and Geopandas.

    Why this beats prompting it yourself

    Manually prompting for scientific workflows often results in agents guessing outdated database parameters or missing critical QC steps like doublet removal. This skill provides a domain-specific map that forces the agent to use verified library patterns and multi-step validation protocols.

    Use cases

    • Identify and rank potential inhibitors for protein targets using chemical databases.
    • Perform quality control and UMAP clustering on 10X single-cell expression data.
    • Annotate pathogenicity and protein impact for variants found in VCF files.
    • Build and compare time-series forecasting models for sensor data with residual analysis.

    Known limitations

    Requires a local Python environment with specific scientific libraries installed. Large-scale genomic or GPU-intensive tasks depend on your local hardware capacity.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Research & Analysis.

    Not a substitute for domain expertise — results still require human validation against source data. Heavier domains (GPU ML, single-cell, docking prep) need extra packages you install and maintain per project.

    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 buying a skill.

    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

    6 uses

    Times used to date

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 2 months ago

    • Free with an Agensi account

    Accounts
    NihEbiEnsemblClinicaltrials
    Software
    Node.jsPython

    Permissions

    Network Access
    Read Files
    Write Files

    Allowed Hosts

    pubchem.ncbi.nlm.nih.gov
    www.ebi.ac.uk
    rest.ensembl.org
    clinicaltrials.gov

    File Scopes

    data/**
    results/**
    *.py
    *.md

    Needs shell and file access to create isolated environments and save intermediate data, figures and reports, plus network access so the agent queries real scientific databases instead of inventing endpoints.

    Listed2 months ago

    What's inside

    Frequently Asked Questions

    This skill provides specialized protocols and domain-specific entry points that prevent the AI from hallucinating outdated API calls and ensure it follows strict QC steps like doublet removal and residual analysis.

    The skill is designed for agents running in Python 3.11+ environments and is compatible with major frameworks like LangChain, AutoGPT, and custom OpenAI Swarms that have access to local code execution.

    You will need to install specific scientific libraries including RDKit, Biopython, and Scanpy; the package includes a detailed requirements file to help you configure your local environment correctly.

    Yes, the skill is pre-configured to handle standard research formats such as FASTA, VCF, and BAM for genomics, as well as molecular structure files for chemical modeling.

    You receive the integration-ready skill package, a comprehensive setup guide, a requirements.txt file for your environment, and access to all future version updates.

    The skill includes built-in logging protocols that track library versions, random seeds, and intermediate data states, ensuring every figure and result can be verified and reproduced.

    No. The base needs only Python 3.11+ and uv or conda; heavier stacks (scanpy, rdkit, pymc) are installed per project only when the task actually requires them, keeping environments light.

    Genomics and sequence work, cheminformatics, proteomics, clinical and translational workflows, ML for science, geospatial analysis, materials and physics computation, plus scientific communication and lab-workflow structuring.

    The playbook enforces fixed seeds, version logging, cached queries and a validation step, but it deliberately requires a human review point before final conclusions — treat outputs as reviewable drafts, not verdicts.

    New in Research & Analysis

    Free