Use when a chromosome model is already defined well enough to choose a conservative simulation route without overcommitting under degeneracy.
Recommend a simulation route only after the model, control, and validation logic are already coherent. The core rule is conservative gating: if the mechanism space is still too degenerate, pause recommendation instead of pretending the software choice is already well-posed.
This skill does not decide the mechanism or build the model. It translates an already-structured model into a conservative simulation paradigm, one or two concrete mainstream engines, and explicit command mappings that can be executed against the current model assumptions.
###[SOFTWARE RECOMMENDATION MANDATE & ANTI-LAZINESS] (CRITICAL)
bond_style fene or hoomd.md.bond.FENE]pair_style lj/cut or pair_style yukawa]fix langevin for Brownian dynamics, or specific Active Matter plugins if needed]Do not use this skill when major degeneracy remains, when the model lacks a control, or when observables and validation have not been defined yet.
Pause recommendation if unresolved degeneracy remains.Recommended Paradigm that matches the model assumptions rather than the most elaborate option available.LAMMPS, HOOMD-blue, OpenMM, or GROMACS.Software Capability Check, Fidelity / Cost Tradeoff, Explicit Solvent / Ion Note, and Limitations & Warnings explicitly.one multiple-choice question per message when one unresolved prerequisite still blocks the recommendation.Do Not Recommend Yet note when the prerequisites are not satisfied.recommendation end node.package-native or package + custom extension.Fidelity / Cost Tradeoff and Explicit Solvent / Ion Note every time.2-4 options plus Other when clarification is still active.Recommended Paradigm and one or two Recommended Engine candidates.Software Capability Check, Fidelity / Cost Tradeoff, Explicit Solvent / Ion Note, and Limitations & Warnings.recommendation end node.stage completeness: did the response include readiness, route, capability check, tradeoff, solvent/ion assumptions, and command mapping?readiness gate: was the gating decision made before tool choice?overcommitted recommendation: did the response oversell a route despite unresolved degeneracy?software verification: were version-sensitive package claims checked against official docs?cost realism: does the chosen fidelity match the actual question?command specificity: did the output name real commands or APIs instead of generic engine classes?pause clarity: if blocked, is the missing prerequisite explicit?question format: did clarification stay in multiple-choice form unless forced otherwise?end-node discipline: did the skill stop at the recommendation end node when the stage was complete?When clarification is still active, use one multiple-choice question per message.
Default question format:
2-4 concrete options.Other as the final option.Every response should use the following structure.
package-native or package + custom extensionUse one or two engine blocks, exactly in this format:
bond_style fene or hoomd.md.bond.FENE]pair_style lj/cut or pair_style yukawa]fix langevin for Brownian dynamics, or specific Active Matter plugins if needed]| Situation | Required response |
|---|---|
| Mechanism and control are still unclear | Pause recommendation and name the missing prerequisite |
| A simple coarse-grained question is being overcomplicated | Choose a conservative paradigm and explain why higher fidelity is unnecessary |
| The user asks for software before model convergence | Say the recommendation is blocked by degeneracy, not by tool availability |
| Explicit solvent or ions are not central to the question | State why implicit treatment is acceptable under the current assumptions |
| Active dynamics are essential | Tie the Recommended Engine to the active ingredients instead of giving a generic package name |
| A package claim is version-sensitive | Verify it in official docs before presenting it as settled |
| Bonded polymer springs dominate the model | Map them to commands such as bond_style fene or hoomd.md.bond.FENE |
| Short-range attraction is sufficient | Map it to pair_style lj/cut or the nearest engine-native equivalent |
| Screened electrostatics matter | Map it to pair_style yukawa or the nearest engine-native equivalent |
| Brownian dynamics is sufficient | Use an explicit dynamics mapping such as fix langevin or the engine-native equivalent |
Fidelity / Cost Tradeoff, which makes the recommendation impossible to evaluate.Explicit Solvent / Ion Note, leaving a major modeling assumption implicit.