Guide · Agent skill
How to use the Flux Analyzer agent skill for metabolic models
Updated · written with AI from the project's own documentation and checked against our catalogue data
The flux-analyzer skill teaches an AI agent to interpret flux balance analysis (FBA) results from a metabolic model, which are predictions of how an organism uses its metabolic network to grow. After following this guide you will know how to install the skill, how to ask the agent for specific analyses, and what files and figures to expect back. It is written for readers who may never have used an agent skill before.
Flux Analyzer for Metabolic Modelsaiming-lab/AutoResearchClaw · MIT Download the skill Quick start
- 1Download the skill ZIP. Get the flux-analyzer ZIP from this page. It contains the SKILL.md file.
- 2Unzip into a skills folder. Extract it to ~/.claude/skills/ for personal use or to a project's .claude/skills/ folder for one project.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, go to Capabilities and upload the skill under Skills.
- 4Install the Python dependencies. Make sure Python, COBRApy (cobra), pandas, numpy and matplotlib are available to the agent.
- 5Place your model and FBA files. Put a COBRA-format model such as my_model.json and an FBA fluxes CSV in your working folder.
- 6Ask for a specific analysis. Request one analysis at a time, such as gene essentiality or a phase plane, naming your files and the outputs you want.
- 7Review the outputs. Check the generated CSV and PDF/PNG files and interpret them as model predictions.
What is an agent skill and what does flux-analyzer do?
An agent skill is a folder containing a SKILL.md file that gives an AI agent instructions for a specific kind of task. When the skill is installed and your request matches its description, the agent follows those instructions.
The flux-analyzer skill covers analysis of FBA flux distributions. Once installed, the agent can write and run Python code using the COBRApy library to find genes and reactions whose removal stops growth, map how growth changes with nutrient supply, sample the range of feasible metabolic flows, summarise activity by pathway, list likely secreted products and produce PDF and PNG figures.
- Gene essentiality: single and double gene deletions, with a growth threshold of 5% of wild-type
- Reaction essentiality: single reaction deletions with the same threshold
- Phenotypic phase plane: growth over a grid of two nutrient uptake rates
- Flux sampling: the OptGP sampler, n = 1000 by default in the example code
- Pathway-level flux aggregation and secretion product prediction
- Publication-style PDF and PNG figures
Check what you need before installing
The skill analyses existing results. It does not build metabolic models from scratch, so you need a COBRA-format model file (the source example uses JSON) and an FBA fluxes CSV before you start.
The skill expects Python with the COBRApy (cobra) library, pandas, numpy and matplotlib installed. The skill runs scripts, so the agent will execute Python code on your machine.
- Python
- COBRApy (cobra)
- pandas, numpy and matplotlib
- A COBRA-format model file, such as my_model.json
- An FBA fluxes CSV with reaction IDs in the index and a flux column
Install the flux-analyzer skill
Download the ZIP for the flux-analyzer skill from this page. The item's files list contains SKILL.md. Unzip the download so that the skill folder sits in one of the locations below.
For a personal install that works across your projects, place the folder in ~/.claude/skills/. To limit it to one project, place it in that project's .claude/skills/ folder.
If you use the Claude apps instead, open Settings, go to Capabilities and upload the skill under Skills.
- Personal: ~/.claude/skills/flux-analyzer/
- Project only: .claude/skills/flux-analyzer/ inside your project
- Claude apps: Settings → Capabilities → Skills
Prepare your model and FBA results
Put your model file and fluxes CSV in the folder where you will work with the agent. The example code loads the model with cobra.io.load_json_model and reads fluxes from a column named flux_mmol_gDW_h, so matching these names avoids edits.
The example code computes wild-type growth from the model's objective before running any analysis. Growth is reported in h^-1.
model = cobra.io.load_json_model('my_model.json')\nfba_fluxes = pd.read_csv('fba_fluxes.csv', index_col=0)['flux_mmol_gDW_h']Ask the agent for an analysis in plain language
You do not need to name the skill. The skill's trigger keywords include metabolic, flux analysis, gene essentiality, synthetic lethality, production envelope, phase plane, flux sampling, secretion, yield and metabolic engineering. Requests that use these terms, or that describe the same analysis, are the ones most likely to load it.
Be specific about the organism or model, the analysis and the output you want. Mention the files you have so the agent can use them.
- Find the essential genes in my_model.json and save a per-gene table as a CSV
- Make a phenotypic phase plane for glucose and oxygen uptake and export it as PDF and PNG
- Run flux sampling with 1000 samples and plot the distributions for central carbon reactions
- Summarise FBA flux by pathway and list the secreted products at optimal growth
Read the outputs and mind the limits
The skill writes named output files, such as gene_essentiality.csv, double_gene_essentiality.csv, reaction_essentiality.csv, flux_samples.csv, pathway_flux_summary.csv and secretion_profile.csv, plus figures such as phenotypic_phase_plane.pdf and pathway_activity.pdf.
Treat the results as model predictions, not experimental evidence. The 5% essentiality cutoff is a common convention and may not suit every organism. Double gene deletions across many genes and 1000-sample flux sampling can be slow and compute-heavy on large genome-scale models, so plan a focused gene set first.
- Start with a focused gene set for double deletions; the example uses the first 50 genes and says to adjust as needed
- Expect long run times for flux sampling on large models
- Check predictions against experimental data before drawing biological conclusions
What people use it for
- Listing genes whose deletion stops growth in a model, saved as gene_essentiality.csv for further study
- Mapping how growth responds to glucose and oxygen uptake to see transitions such as aerobic growth versus fermentation
- Sampling feasible flux distributions to find central metabolism reactions with wide or narrow ranges
- Summarising FBA activity by metabolic subsystem to produce a pathway bar chart for a report
- Listing likely secreted by-products at optimal growth as a starting point for metabolic engineering
Frequently asked questions
Is the flux-analyzer skill free to use?
The listing states that the skill is released under the MIT licence. Check the licence file in the source repository for the full terms before reuse.
What do I need before I can use the flux-analyzer skill?
You need Python with COBRApy, pandas, numpy and matplotlib, plus a COBRA-format model file such as JSON and an FBA fluxes CSV. The skill analyses existing results and does not build models from scratch.
How do I install an agent skill like flux-analyzer?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings, then Capabilities, then Skills.
Can I use flux-analyzer without a metabolic model?
No. The skill works on existing COBRA-compatible models and FBA output. It does not create models, so you need those files first.
Does flux-analyzer need an internet connection?
The listing says the skill does not use the network. It does run Python scripts on your machine, so review the code the agent produces before running it on sensitive data.
How is flux-analyzer different from running COBRApy by hand?
The skill gives the agent a set of worked analysis patterns, including essentiality, phase planes, flux sampling, pathway summaries and secretion profiles, with named outputs and figures. You still need COBRApy and a model, and the results still need expert interpretation.
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
Use flux-analyzer if you already have a COBRA-compatible metabolic model and FBA output, and you want an agent to run standard analyses and produce figures quickly. Pick something else if you need to build models from scratch, or if you need clinical or safety-critical conclusions, since the outputs are model predictions rather than experimental evidence.
Download the skill