Calculate Activation Barriers with MLIPs: chem-neb-barrier

Where this fits

A researcher asks an agent to compute the energy barrier for lithium diffusion in a cathode material. Without a dedicated skill, the agent would need to manually script NEB simulations, handle MLIP selection, and parse outputs. This skill packages that workflow into a single callable tool, letting the agent focus on the scientific question rather than the mechanics.

Why agents benefit

• Automates NEB setup: Generates intermediate images between start and end structures with a single command.
• Auto-detects periodicity: Switches seamlessly between solid-state (periodic) and molecular (non-periodic) systems.
• Integrates with MACE MLIPs: No need to write custom potential wrappers—just specify --model_type mace.
• Returns actionable results: Outputs the barrier height and the transition state path, ready for downstream reasoning.
• Handles convergence: Uses --fmax to ensure accurate transition states, reducing agent debugging loops.

Practical scenarios

Solid-State Diffusion

Compute vacancy migration barriers in battery materials. The agent can feed two relaxed CIF structures and get the activation energy for lithium hopping, directly informing ionic conductivity estimates.

Molecular Reaction Pathways

Find the activation energy for a chemical reaction in vacuum. The skill detects non-periodic boundary conditions and adjusts the NEB algorithm accordingly, making it suitable for organic chemistry transition states.

High-Throughput Screening

Run NEB for hundreds of candidate materials. An agent orchestrates the workflow, calls this skill per pair of endpoints, and collects barriers for a machine-learning model training set.

Add it to your agent workflow

Install the skill with:

npx skills add learningmatter-mit/AtomisticSkills --skill chem-neb-barrier

Then use it in your agent's toolchain. For a periodic material (e.g., a CIF file pair):

python .agents/skills/chem-neb-barrier/scripts/calculate_barrier.py \
    --start_structure start.cif \
    --end_structure end.cif \
    --model_type mace \
    --n_images 5 \
    --fmax 0.05 \
    --output_dir results/

The script auto-detects periodicity, so the same command works for molecular structures (e.g., XYZ files) without extra flags.

Compare before adopting

Before integrating this skill, compare it against:

• Manual scripting: Writing your own NEB loop with ASE or LAMMPS—more flexible but slower to implement.
• Other NEB tools: Standalone software like VTST or CatMAP, which may require separate MLIP integration.
• Maintenance freshness: Check the repo's commit history and issue tracker to ensure it stays compatible with newer MACE versions.
• Workflow fit: Verify that the skill's input/output format matches your agent's existing file handling and logging conventions.

Why it is worth tracking

This skill is part of the AtomisticSkills repository (161 stars) from MIT’s Learning Matter group, indicating active research use and community adoption. It abstracts complex physics into a clean CLI, making it a strong candidate for agents that need to reason about reaction energetics. Evaluate it when your agent must answer “what is the barrier?” questions reliably and quickly. For more details, see the skill page at https://www.openagentskill.com/skills/learningmatter-mit-chem-neb-barrier.