// Skills for opening and driving agent-controllable visualization components in the Pantheon UI sidebar — interactive viewers the agent can open, control, and read back. Viewers: Vitessce (spatial / single- cell omics), Viv (bioimage / microscopy), plus agent-generated apps.
| id | live_view_index |
| name | LiveView Skills Index |
| description | Skills for opening and driving agent-controllable visualization components in the Pantheon UI sidebar — interactive viewers the agent can open, control, and read back. Viewers: Vitessce (spatial / single- cell omics), Viv (bioimage / microscopy), plus agent-generated apps. |
| tags | ["live-view","visualization","vitessce","viv","bioimage","spatial","single-cell","interactive"] |
A LiveView is an interactive component the agent opens in the Pantheon
UI's right sidebar, then drives and observes through the live_view tools.
Unlike a static plot image, a LiveView is live: the agent changes its state
and the user sees it update; the user interacts with it and the agent reads
the result back.
Load the relevant skill file before building a visualization.
Architecture — only the LiveView SDK runtime is built in. Every viewer
is a plugin: a folder skills/live_view/<name>/ holding <name>.md
(this guide), adapter.js (a setup(lv, root) module), and an optional
demo.json. open_live_view either resolves a named viewer plugin
(view_type="vitessce") or loads an agent-generated component
(view_type="custom" + module_url). Adding a viewer = dropping a new
<name>/ folder here; no app code changes.
Open a Vitessce browser to explore spatial transcriptomics, single-cell, and microscopy-imaging datasets: spatial scatterplots, gene-expression coloring, heatmaps, cell-set selection, image layers.
Skill file: vitessce/vitessce.md
When to use:
Open a Viv viewer for high-resolution, multiplexed bioimaging — OME-TIFF and OME-Zarr (OME-NGFF): multichannel fluorescence, microscopy, IF/IMC/ CODEX, whole-slide images. Channel colors, contrast, pan/zoom, overview.
Skill file: viv/viv.md
When to use:
serve_local_data)Open a Mol* viewer for 3D macromolecular structures — proteins, nucleic
acids, complexes — from the RCSB PDB, the AlphaFold DB, or a local
.pdb / .cif file. Rotate, zoom, inspect; AlphaFold models colour by
pLDDT.
Skill file: molstar/molstar.md
When to use:
.pdb / .cif structure file (see also the structural_biology skill
for obtaining / predicting structures)Open an IGV.js genome browser to view BAM/CRAM alignments, VCF variants, BED/GFF annotations, bigWig coverage — on a reference genome (hg38, mm10, custom FASTA, …). Pan, zoom, jump to a gene or locus.
Skill file: igv/igv.md
When to use:
Open a Gosling.js view for designed genomic visualisations — circular ideograms, multi-track / multi-sample layouts, comparative dual-genome views, custom encodings. Driven by a declarative JSON spec.
Skill file: gosling/gosling.md
When to use:
Open a Cytoscape.js view for interactive networks — protein-protein interactions, signalling / metabolic / regulatory pathways, ontology graphs. Nodes + edges as JSON, built-in layouts (cose, breadthfirst, circle, dagre, ...), CSS-like stylesheet by selectors.
Skill file: cytoscape/cytoscape.md
When to use:
Open a multiple-sequence-alignment view (EBI Nightingale's
<nightingale-msa>) for protein or DNA alignments. Standard colour
schemes (clustal, taylor, hydro, zappo, ...), configurable tile sizes.
Skill file: msa/msa.md
When to use:
phylotree for tree + alignment side-by-sideRender 2D depictions of small molecules from SMILES / MOL block using
RDKit-JS (WebAssembly build). Complements molstar (3D macromolecules)
— RDKit is the canonical 2D view for drugs, metabolites, organics.
Skill file: rdkit/rdkit.md
When to use:
Open a phylotree.js view for an interactive phylogenetic tree from a Newick string. Linear or radial layout, branch-length-scaled; rerooting, ladderise, clade collapse built in.
Skill file: phylotree/phylotree.md
When to use:
msa for tree + alignment side-by-sideWrite your own interactive component with the LiveView SDK when no existing viewer fits — a bespoke dashboard, custom plot, or tailored data view that the agent can still open, drive, and observe.
Skill file: live-view-app.md
When to use:
live_view tools| Tool | Purpose |
|---|---|
open_live_view(view_type, title, state, module_url?) | Open a viewer plugin (e.g. view_type="vitessce") or a custom component (view_type="custom" + module_url); returns view_id |
serve_local_data(path) | Expose a workspace file/dir over HTTP+CORS; returns a fetchable URL |
live_view_update(view_id, patch) | Deep-merge a partial-state patch (drive it) |
live_view_set_state(view_id, state) | Replace the whole state |
live_view_get_state(view_id) | Read state, status, and diagnostics — incl. the user's own edits |
live_view_call(view_id, action, args) | Invoke a component-defined action |
live_view_screenshot(view_id) | Render the view to an image — observe_images it to see it |
list_live_views() / close_live_view(view_id) | List / close |
Workflow: open_live_view → verify (live_view_get_state for
diagnostics, live_view_screenshot to see it — status: ready does NOT
mean it rendered correctly) → drive with live_view_update →
live_view_get_state before the next move. Never treat reading back your
own live_view_update value as verification.
Skills for rare disease case support: ontology-first normalization/retrieval and the clinical genetics consult report format contract (structure + theme). Load the relevant skill file when performing the matching task.
Scenario router for paper-writing tasks. Use after root triage to select paper submission, journal article, conference paper, grant proposal, lab report, group report, talk/workshop, or revision-response behavior.
Routing and workflow skill family for paper-writing tasks. Covers manuscript drafting, journal and conference papers, grant proposals, lab reports, group-meeting reports, talks, workshop notes, reviewer rebuttals, academic HTML/PDF/LaTeX output with editable-block contracts, citation grounding, evidence checking, and pre-submission quality gates.
Workflow phases for paper-writing tasks: triage, material inventory, research question, literature review, paper outline, data analysis summary, figure storyline, reader testing, and finalize packet. Each phase is a short contract — read the relevant rows for the current task only.
Section-level writing skill for evidence-bound academic prose: IMRaD papers, grants, reports, talks, and response letters. Indexes section-specific templates (abstract, introduction, method, results, discussion) and pre-submission quality protocols (claim-evidence, reviewer rubric, response letter).
Skills for querying and downloading data from genomic, transcriptomic, 3D-genome, and cancer-genomics databases. Covers programmatic access to public repositories, gene annotation, sequence retrieval, processed functional-genomics tracks, Hi-C / Micro-C contact matrices, TCGA-style cohorts, and large-scale single-cell data.