How to use the Cantera Ignition Delay Calculator agent skill

What an agent skill is and what this one does

An agent skill is a folder of instructions, reference notes and scripts that an AI agent can load when a task matches its description. This skill is named cantera. Its description says it runs Cantera homogeneous chemical reactors and evaluates ignition delay, with mechanism provenance, conservation checks and numerical refinement.

The skill covers a closed, adiabatic, ideal-gas reactor at constant volume or constant pressure. It measures how long a gas mixture takes to ignite, defined as the moment of fastest temperature rise. It then re-runs the case with finer output, tighter solver settings and a longer time horizon to check whether the answer is numerically stable, and it checks that mass and energy are conserved.

• Best for combustion researchers or engineering students comparing ignition delay under controlled conditions.
• Best for developers who need a reproducible, scripted Cantera calculation with saved reports.
• Best for analysts checking whether a kinetic mechanism reproduces a simple ignition case numerically.
• Not for flame speed or flame structure questions. The skill is explicitly not a flame solver.
• Not for safety-critical decisions such as explosion limits or process design.

How to install the Cantera skill

Option 1 is to download the ZIP file from this page and unzip it into your personal skills folder, ~/.claude/skills/. To make the skill available only inside one project, unzip it into that project's .claude/skills/ folder instead.

Option 2 is to upload the ZIP in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the file there.

Installing the skill does not install the scientific software it uses. The first run needs network access to download Cantera, NumPy and uv.

• Requirements: Python in the range the skill lists, Cantera at the release the skill is tested against, and NumPy.
• uv is used in the example command to install dependencies for that run.
• No API keys or accounts are needed.

Ask for an ignition delay calculation in plain language

You do not need to name the skill or write code. Describe the mixture, the starting temperature and pressure, and whether the reactor is constant volume or constant pressure. The skill is meant for closed, adiabatic, homogeneous ideal-gas cases with a known kinetic mechanism.

Good requests name the physical setup. Vague requests such as 'how fast does gas burn' do not match the skill's scope. If you have a specific mechanism file, mention it, because the skill needs mechanisms available as Cantera YAML files.

• Calculate the ignition delay of a 2:1:7 H2:O2:Ar mixture at 1000 K and 101325 Pa in a constant-volume reactor.
• Compare constant-volume and constant-pressure ignition delay for the same hydrogen mixture and report the mechanism hash.
• Check whether my mechanism reproduces this simple ignition case numerically, and list any refinement changes.
• Run the bundled hydrogen example and save the report to a new output folder.

Run the bundled example yourself

The skill includes a tested example configuration. From the collection root, the example command below runs the helper script on the bundled hydrogen case and writes results to a folder named hydrogen-result. Use a new output folder for each run.

The skill documents that the 1000 K, 101325 Pa, H2:O2:Ar = 2:1:7 constant-volume example gives about 0.313 ms using its stated delay definition. These are package regression values for checking that the setup works, not experimental data.

Read the outputs and checks

Each run writes a JSON report and four CSV files of time histories, one for each refinement run. The report lists input settings, package versions, configuration and mechanism hashes, the normalized starting composition, the four delay estimates, numerical changes and conservation results.

The delay is reported only if the temperature rise reaches the configured minimum and the peak is well inside the simulation time. Otherwise the delay is null and a status explains why. A successful run shows that the numerics worked. It does not show that the mechanism is valid for your fuel, pressure or diluent.

• report.json: settings, hashes, delay estimates, refinement changes and conservation checks.
• baseline.csv, finer_output.csv, tighter_solver.csv and longer_horizon.csv: time, temperature, pressure, volume, mass, energy, enthalpy and requested species mole fractions.
• mechanism.yaml: a Cantera-written snapshot of the loaded mechanism.

Safety notes: scripts and network access

This skill runs a Python script on your machine and uses the network for the initial package installation. The source notes that simulations run locally, with no credentials. Review the script and the configuration file before running them, especially if the configuration came from someone else.

Custom mechanisms need their original dependency files kept alongside the generated snapshot so the calculation can be replayed later.

• Runs scripts: yes.
• Uses network: yes, for the initial package installation.
• No API keys or accounts.

Tips for reliable results

Check that the mechanism's validated condition range covers your case before trusting the number. The bundled hydrogen example is a demonstration, not a recommendation for every hydrogen experiment.

If the delay changes materially between refinement runs, or the peak sits close to the end of the simulation, refine again with a longer horizon or finer output spacing. Always report the delay together with the delay definition, output spacing and scientific limits.

• Set the time horizon long enough to see the temperature rise and the drop after the peak.
• Choose output spacing fine enough to locate the peak.
• Use a new output directory for each run.
• Compare temperature and species histories with the ignition definition you actually care about.