---
name: biosimulant
title: Biosimulant Skill Router
description: Route Biosimulant tasks to the right public skill for CLI use, lab authoring, scientific publishing, and runtime-specific model work.
version: 2026.05.23
tags: [biosimulant, router, labs, skills]
audience: Claude Code, Cursor, Continue, Aider
recommended: true
---

# Biosimulant Skill Router

Use this skill first when a user asks an agent to create, clean, validate, document, run, or publish a Biosimulant model
or lab. It routes the task to the smallest specific skill that covers the work.

## What Biosimulant Is

Biosimulant is a platform for packaging scientific models as runnable labs. A lab combines one or more `BioModule`
models, a `lab.yaml` composition, explicit inputs and outputs, runtime settings, visualisations, README evidence, and
optional Hub publication metadata. Publish-worthy labs should be source-faithful, runnable, documented, and visually
useful.

Common lab families:

- mechanistic simulators from SBML or CellML source models
- ONNX model artifacts for imaging, sequence, embedding, segmentation, or reconstruction workflows
- custom Python or mixed-source models with documented equations, parameters, and assumptions

## Skill Selection

- Use `biosimulant-cli.md` for CLI commands, auth, run inspection, local serving, packages, and Hub publishing.
- Use `biosimulant-lab-authoring.md` when creating or restructuring a lab or model repository.
- Use `biosimulant-scientific-publishing.md` when a lab must be publish-worthy, source-faithful, documented, and
  public-ready.
- Use `biosimulant-sbml.md` for SBML-backed labs and `TelluriumSBMLBioModule`.
- Use `biosimulant-cellml.md` for CellML-backed labs and `LibCellMLBioModule`.
- Use `biosimulant-onnx.md` for ONNX-backed labs, graph-contract validation, tensor inputs, and model-card caveats.
- Use `biosimulant-custom-models.md` for custom Python, VirtualCell-style, handwritten, or mixed-source labs.
- Use `biosimulant-readme-publish.md` for screenshot README generation, Hub publishing, and public completed-run checks.

Use multiple skills when the task spans phases. Example: cleaning a CellML repo for publication should use
`biosimulant-lab-authoring.md`, `biosimulant-cellml.md`, `biosimulant-scientific-publishing.md`, and
`biosimulant-readme-publish.md`.

## Global Rules

- Prefer source-tree labs with explicit manifests and versioned source artifacts.
- Preserve source scientific artifacts; do not manually rewrite equations or parameters to make a lab easier to run.
- Public inputs and outputs must be friendly, traceable, and deliberate.
- Visualisations must use actual runtime outputs and must not be empty.
- READMEs must explain provenance, assumptions, input/output mappings, visuals, validation, and caveats.
- Public Hub publication is complete only after `biosimulant hub labs get <hub_id> --json` reports
  `is_public == true` and `completed_runs >= 1`.
- If a lab cannot be made source-faithful and runnable without inventing unsupported science, do not publish it.

## Minimal Workflow

1. Identify source type: SBML, CellML, ONNX, custom Python, or mixed.
2. Read `lab.yaml`, `models/core/model.yaml`, source artifacts, wrappers, README, and tests.
3. Create or update a per-lab plan before bulk edits.
4. Keep or orphan based on source-faithful + runnable evidence.
5. Make ports, visuals, and README user-facing without weakening traceability.
6. Validate manifests, entrypoints, runtime outputs, visuals, and README assets.
7. Run at least one successful local run before publishing.
8. Publish only when the lab passes the public Hub run gate.

## Example Routing

- "Create a BioModels epidemiology lab from SBML" -> use `biosimulant-lab-authoring.md`,
  `biosimulant-sbml.md`, `biosimulant-scientific-publishing.md`, and `biosimulant-readme-publish.md`.
- "Clean 400 CellML physiology labs for release" -> use `biosimulant-lab-authoring.md`,
  `biosimulant-cellml.md`, `biosimulant-scientific-publishing.md`, and `biosimulant-readme-publish.md`.
- "Publish ONNX segmentation labs with better visuals" -> use `biosimulant-onnx.md`,
  `biosimulant-scientific-publishing.md`, `biosimulant-cli.md`, and `biosimulant-readme-publish.md`.