Skip to content
Get packs

Guide · Agent skill

How to use the Glycoengineering Assistant agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The glycoengineering skill teaches an AI agent how to look at a protein sequence for places where sugar chains (glycans) are likely to attach, and how to plan careful changes such as adding or removing those sites. It is aimed at protein engineers, antibody scientists and students who want a structured, checkable workflow rather than a loose guess. By the end of this guide you will know what the skill does, how to install it, how to phrase a request so the agent uses it, and what its limits are.

Glycoengineering AssistantK-Dense-AI/scientific-agent-skills · MIT

Quick start

  1. 1Download the ZIP. Download the glycoengineering skill ZIP from this page.
  2. 2Unzip into a skills folder. Unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
  3. 3Or upload in the Claude apps. In the Claude apps, open Settings, go to Capabilities and upload the skill under Skills.
  4. 4Check requirements. Confirm Python 3.10+ is available for the local helper, and install requests if you plan to use the GlyTouCan lookup.
  5. 5Ask with a plain request. Describe the protein, the sequence and your goal, such as scanning for N-glycosylation sites before a mutation.
  6. 6Verify with the prediction services. Use the official NetNGlyc or NetOGlyc web forms to cross-check candidates, keeping the input and output together.

What the glycoengineering skill makes the agent do

An agent skill is a folder of instructions and optional scripts that an AI agent loads when a task matches the skill's description. This one covers protein glycosylation. It tells the agent to scan sequences for canonical N-glycosylation sites (the pattern N-X-S/T, where X is not proline), to describe serine- and threonine-rich regions that may be mucin-type O-glycosylation candidates, and to check proposed mutations for sites that are lost or created.

The skill also asks the agent to keep four kinds of result separate: the sequence motif, a computational prediction, experimental evidence that a site is occupied, and the glycan's structure or composition. A motif alone does not prove a glycan is attached.

  • Finds N-X-S/T sequons, keeping overlapping sites such as NNST
  • Flags serine/threonine-rich windows as descriptive O-GalNAc candidates
  • Rescans an engineered sequence to show which motifs were lost or created
  • Prepares workflows for NetNGlyc, NetOGlyc and GlycoSHIELD
  • Looks up glycan descriptions in GlyTouCan

Install the skill from the ZIP or the Claude apps

Download the ZIP of the skill from this page, then choose one of two routes. For a personal setup, unzip it into ~/.claude/skills/ so the glycoengineering folder sits there. For a single project, unzip it into the project's .claude/skills/ folder instead.

If you use the Claude apps rather than a local folder, open Settings, go to Capabilities, and upload the skill under Skills.

  • Personal: ~/.claude/skills/ (unzip here)
  • Project only: .claude/skills/ inside your project folder
  • Claude apps: Settings → Capabilities → Skills, then upload

Trigger the skill with natural requests

You do not need a special command. The skill loads when your request matches its description, so mention the task in plain language. Good triggers include glycoprotein engineering, antibody Fc glycosylation, glycan shielding, and interpreting site-specific glycoproteomics data.

Include the sequence and the goal. The agent will ask for missing details such as the isoform, the expression host or whether a signal peptide is present, because the skill asks for them before drawing conclusions.

  • "Scan this antibody Fc sequence for N-glycosylation sites and list their positions."
  • "I want to remove the glycosite at Asn297 in this heavy chain. Check what else changes."
  • "Triage this mucin-like sequence for O-GalNAc candidates before I run NetOGlyc."
  • "Help me plan a glycan shielding approach for this structure with GlycoSHIELD."

Check what the skill needs before you start

The local sequence helper is a Python script that uses only the standard library and needs Python 3.10 or newer. The GlyTouCan lookup script needs the requests library and network access. The DTU predictors (NetNGlyc and NetOGlyc) are used through a web browser, not an API.

GlycoSHIELD is optional. It needs a separate source installation, glycan libraries and GROMACS if you want solvent-accessible surface area (SASA) analysis.

  • Python 3.10+ for the local sequence helper (standard library only)
  • requests library and network access for the GlyTouCan lookup
  • A web browser for NetNGlyc and NetOGlyc
  • Optional: GlycoSHIELD source install, glycan libraries and GROMACS

Understand the scripts and network use

This skill runs scripts and uses the network. The sequence helper scripts/glycoengineering_tools.py works offline on the sequence you give it. The scripts/glytoucan_lookup.py script queries public GlyTouCan services. The skill does not submit sequences to prediction servers automatically; it uses the official web forms.

Review scripts before running them, and avoid sending confidential or unpublished sequences to public web services unless your organisation allows it. The reference files cover the glycan databases and GlycoSHIELD workflow in more detail.

  • scripts/glycoengineering_tools.py: local, standard-library sequence analysis
  • scripts/glytoucan_lookup.py: public GlyTouCan lookup over the network
  • references/glycan_databases.md and references/glycoshield.md: background and source details
  • SKILL.md: the workflow the agent follows

Read the helper output correctly

The helper accepts a raw canonical 20-amino-acid sequence, normalises whitespace and case, and rejects FASTA headers, gaps and unknown residues. Positions are 1-based in the normalised submitted sequence. Parse FASTA records first, because removing unknown residues would shift coordinates.

An N-to-Q change can alter protein behaviour even without a glycan effect, and the helper does not predict folding, function or tolerability. Treat every output as a checklist item for experimental follow-up.

  • Motif present does not mean occupied
  • S/T density is a fraction of residues, not an O-glycosylation probability
  • Validate occupancy experimentally before any engineering decision
sequence = normalize_sequence("NNST") sites = find_n_glycosylation_sequons(sequence) assert [site["position"] for site in sites] == [1, 2]

What people use it for

  • Listing N-glycosylation candidates in an antibody Fc region before planning a change
  • Checking whether a proposed mutation removes one site and creates another overlapping site
  • Triaging a mucin-like sequence for O-GalNAc candidates before running NetOGlyc
  • Looking up a glycan accession in GlyTouCan to see its WURCS description
  • Preparing a GlycoSHIELD glycan shielding workflow for a protein structure

Frequently asked questions

How do I install the glycoengineering agent skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings, then Capabilities, then Skills.

What does the glycoengineering skill require?

The local helper needs Python 3.10 or newer and uses only the standard library. The GlyTouCan lookup needs the requests library and network access, and the NetNGlyc and NetOGlyc predictors are used through a web browser. GlycoSHIELD and GROMACS are optional.

Is the glycoengineering skill free to use?

The skill is listed under the MIT licence. Check the licence file in the download for the full terms before using it in your own work.

Does the skill submit sequences to NetNGlyc or NetOGlyc automatically?

No. The skill uses the official web forms and provides no stable API for job submission, so you run the predictions yourself in a browser.

Can the skill prove that a glycosite is occupied?

No. A sequon motif or a prediction does not establish occupancy. Occupancy and glycoforms need experimental validation, such as glycan analysis of the protein.

How does this compare with running NetNGlyc or NetOGlyc directly?

The skill gives the agent a structured checklist for sequon scanning, mutation rescans and evidence separation, and it prepares the prediction steps. The DTU web services themselves still do the prediction, so use the skill to plan and the official forms to predict.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill suits protein engineers, antibody scientists and students who want a disciplined checklist for N-glycosylation sequons and O-GalNAc candidates before running experiments or prediction services. It is not suitable for clinical or regulatory decisions about biosimilars or therapeutics, or for predicting whether a mutant folds, works or is safe.

Download the skill