Guide · Agent skill
How to use the MFA Pipeline Orchestrator agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The MFA Pipeline Orchestrator is an agent skill for Claude that coordinates a multi-step metabolic flux analysis workflow, from loading a genome-scale metabolic model through to phenotype prediction and publication-style figures. This guide explains what an agent skill is, how to install this one, how to ask for a run in plain language, and where its limits are. By the end you will know what inputs to provide, what files to expect, and when this skill is the right tool.
MFA Pipeline Orchestratoraiming-lab/AutoResearchClaw · MIT Download the skill Quick start
- 1Download the ZIP. Get the skill ZIP from the page and unzip it to produce the skill folder containing SKILL.md.
- 2Place the skill folder. Copy it to ~/.claude/skills/ for personal use, or to .claude/skills/ in a project, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Install the required sub-agents. Make sure model-builder, fba-runner, flux-analyzer and metabolic-pheno-analyzer from the mfa-agent set are available, since they are not included in this file.
- 4Prepare your inputs. Decide on the organism or BIGG model ID or custom reactions, the growth condition, the objective, any knockouts and the analyses you want.
- 5Make a natural request. Describe the organism, model ID or reaction list and ask for end-to-end metabolic modelling in plain language.
- 6Check the progress files. Open the progress/ markdown files to see each step's PASS or FAIL status and the paths to outputs.
- 7Resume after a failure. If a step fails, ask the agent to resume from it. Completed steps are not re-run.
What is an agent skill, and what does this one make the agent do?
An agent skill is a folder containing a SKILL.md file with instructions that tell an AI agent how to handle a particular kind of task. When a request matches the skill's description, the agent follows those instructions instead of improvising.
This skill covers an end-to-end metabolic modelling run. It is triggered when you provide an organism name, a BIGG model ID, or a custom reaction list and want the whole process run automatically. The skill sets out four stages: model-builder produces a model file and validation report; fba-runner produces flux CSVs and scan summaries; flux-analyzer produces an essentiality table and a phase plane figure; and metabolic-pheno-analyzer produces PDF figures and a yield table.
The orchestrator itself does not perform the flux calculations. It hands each stage to a specialised sub-agent and records progress in markdown files, so if one step fails you can resume from that step rather than starting over.
What do you need before you install?
The skill depends on sub-agents named model-builder, fba-runner, flux-analyzer and metabolic-pheno-analyzer from the mfa-agent set. These are not included in this file, so the skill will not run end-to-end without them.
The skill also assumes underlying tools such as COBRApy and access to BIGG model data. The source file does not specify how these should be installed or configured, so check the mfa-agent set and your own environment for those details.
Install the skill from the ZIP download
Download the ZIP for this skill from the page you are reading. Unzip it and place the resulting skill folder in one of two locations. For a personal install available across your projects, use ~/.claude/skills/. For a single project, use the .claude/skills/ folder inside that project.
If you use the Claude apps instead, open Settings, go to Capabilities, and upload the skill under Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project folder
- Claude apps: Settings → Capabilities → Skills
How to trigger the skill with a natural request
The skill triggers when your request includes an organism name, a BIGG model ID, or a custom reaction list, and asks for end-to-end metabolic modelling to run automatically. Include the details the workflow needs: the model source, the growth condition (aerobic or anaerobic, carbon source and concentration), the objective (biomass or a product), any gene knockouts, and what analyses you want.
The source lists these analysis goals: essentiality, phase plane, yield optimisation, and wild-type versus mutant comparison. If you want a theoretical yield for a product, name the target product.
- Model source: a BIGG ID or your custom reactions
- Condition: aerobic or anaerobic, carbon source, concentration
- Objective: biomass or a target product
- Knockouts: gene names to apply, if any
- Analyses: essentiality, phase plane, yield optimisation, WT vs. mutant
Example requests
Run a full metabolic flux analysis on a BIGG model for an organism, with an aerobic glucose condition and a knockout screen, and produce publication figures and a yield table for ethanol.
Using this custom reaction list, build the model, run FBA with pFBA and FVA, and identify essential genes and phase plane optima for glucose and oxygen.
The source does not include sample outputs, so treat these as request patterns rather than verified results.
What files does a run produce?
The skill tracks progress in a progress/ folder. Each completed step writes a markdown file with a status of PASS or FAIL. Step 1 writes progress/step1_metabolic_model.md with the model file and validation details. Step 2 writes progress/step2_fba_simulation.md with the flux CSVs. Step 3 writes progress/step3_flux_analysis.md with the phase plane and engineering targets. Step 4 writes progress/step4_metabolic_phenotype.md with the maximum theoretical yield and figure paths.
All file paths are relative to the working directory. If a step fails, the orchestrator checks these progress files and resumes from the failed step rather than re-running completed ones. Total sub-agent retries are capped at 10 across all steps.
- models/<Model>.json and a validation report
- simulations/fba_fluxes.csv and scan_summary.json
- analysis/essentiality.csv and analysis/phase_plane.png
- output/figures/*.pdf and a yield table
progress/step1_metabolic_model.md
progress/step2_fba_simulation.md
progress/step3_flux_analysis.md
progress/step4_metabolic_phenotype.mdSafety notes: scripts and network access
The skill file does not run scripts itself, and according to the listing it does not use the network. Its computation is delegated to the sub-agents, so the behaviour of those sub-agents and any tools they call is outside what this file specifies. Review them before running on sensitive data.
The outputs are model predictions based on simplified assumptions. The skill does not validate biological accuracy, so results need expert interpretation. Do not use them to make clinical or medical decisions.
- Runs scripts: no (per the listing)
- Uses network: no (per the listing)
- Sub-agents and underlying tools are not included in this file
What people use it for
- Screening a known organism's genome-scale model for genes whose deletion stops growth.
- Estimating the theoretical maximum yield of a product such as ethanol before planning lab work.
- Comparing wild-type and knockout growth under different carbon sources or oxygen levels.
- Generating a phase plane figure for two nutrients to see growth limits.
- Producing PDF figures and a yield table for a metabolic engineering report.
Frequently asked questions
How do I install the MFA Pipeline Orchestrator skill?
Download the ZIP from the page, unzip it, and place the skill folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder. In the Claude apps you can instead upload it under Settings → Capabilities → Skills.
Is the MFA Pipeline Orchestrator free to use?
The skill is listed under the MIT licence. The source file does not state pricing for the sub-agents or other tools it relies on, so check those separately.
What do I need to run this skill?
You need the sub-agents model-builder, fba-runner, flux-analyzer and metabolic-pheno-analyzer from the mfa-agent set, which are not included in this file. The skill also assumes tools such as COBRApy and access to BIGG model data.
Does the skill run scripts or use the internet?
According to the listing, it runs no scripts and uses no network directly. The sub-agents and their underlying tools are not specified in this file, so review them before use.
What can I do if a step fails?
Check the progress markdown files to see which step failed. The orchestrator resumes from that step without re-running completed ones, with a maximum of 10 sub-agent retries across all steps.
How does this compare with running COBRApy myself?
The skill chains several analyses together automatically and tracks progress, which suits a full pipeline. For a single quick calculation, the listing says the multi-agent overhead is not worth it, so running COBRApy directly is simpler.
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 metabolic engineering researchers and students who want an automated first pass at flux balance analysis on a known organism, and computational biologists who already use COBRApy and BIGG models. It is not suited to clinical decisions, to users without a working understanding of metabolic models, or to single quick calculations, and you should pick a simpler tool for those.
Download the skill