How Skills work in a project
Skills are per project, not global. Each project has its own attached set with a toggle for each one. The agent reads only the active Skills.1
Open the Skills panel
Click the wrench icon in the project’s right sidebar. A panel opens listing every Skill attached to this project, with a toggle next to each.
2
Read the active count
The header shows how many Skills are enabled out of the total attached — for example, “4 of 7 enabled”. The agent applies only the enabled Skills.
3
Add Skills from the catalogue
Click + Add to open the catalogue. Search or browse by category. Each result is a card with a one-line description; click for the full text.
4
Toggle as you go
Switching a toggle takes effect on the next prompt. The agent will show a Shell step like “Enable pydeseq2 Skill” in the Steps list when it consults it — proof the toggle worked.
The catalogue
Galagos ships with 250+ prebuilt Skills covering the most common bioinformatics tools and workflows. Categories include:- Genomics and pipelines — alignment, variant calling, quantification, QC.
- RNA-seq and single-cell — DESeq2, pydeseq2, edgeR, Scanpy, Seurat.
- Protein structure and design — AlphaFold, AlphaFold-Database, ColabFold, RoseTTAFold.
- Chemistry and small molecules — RDKit, AutoDock Vina, ChEMBL.
- Proteomics and metabolomics — peak-finding, identification, quantification.
- Drug discovery — target search, docking, screening.
- Clinical and medical — clinical-variant interpretation, evidence linking.
Browse pre-built Skills
Examples organised by category, with descriptions.
Workspaces
How workspace owners pin or require Skills across every project.
Workspace-level Skills
In a team workspace, owners and admins can configure Skills at the workspace level:- Pinned Skills become defaults — they are pre-attached when a member creates a new project.
- Required Skills cannot be toggled off — useful for QC thresholds, citation policy, or privacy rules that apply across every project.
Save a conversation as a Skill
After you’ve run an analysis you want to repeat, save the conversation itself as a Skill. The agent extracts the approach, the parameters, and the interpretation logic from the chat and turns them into a reusable Skill. You can do this two ways:- From the chat — after a successful run, the agent often offers “Save this as a Skill?” as a suggestion. Click it.
- Manually — open the Skills panel and click Create. Describe in plain English what the Skill should do; the agent drafts it and asks you to confirm.
Custom Skills in plain English
A custom Skill is just a short brief. Writing examples:- “Whenever I upload gene expression data, log-transform counts, filter out genes with zero reads in more than half the samples, and include both fold change and adjusted p-value in the output.”
- “For every phylogenetic tree, root on the outgroup I tell you, colour branches by host species, and scale branch lengths to substitutions per site.”
- “When I share a paper, extract the methods, the figures, and the supplementary tables. I don’t need the introduction.”
A few tips
- Brief like a new lab member, not a computer. Tell the Skill why and when, not just what.
- One sentence is enough to start. You can refine later.
- Layer two or three per project. Avoid attaching a Skill just in case — it changes how the agent works on every prompt.
- Update Skills when your approach changes. A six-month-old Skill that says “use limma-voom” may not be what you want next month. Edit it, or replace it.
Related pages
Pre-built Skills
Examples from each catalogue category.
Custom Skills
Write a Skill from chat in a couple of minutes.
Workspaces
Pin or require Skills at the workspace level.
Memory
Long-term facts about how the team works, applied automatically.

