agentleFS
SFETNISFETNI/Deep-Matter-Chem-Skills/skills/continuum-modeling/creep-mechanics-workflow/SKILL.md

creep-mechanics-workflow

Skill10 starsChanged 6 months agoMITView on GitHub

Select, calibrate, implement, and validate creep deformation and rupture models for materials and components under sustained thermomechanical loading. The workflow separates rate prediction, transient deformation, stress redistribution, damage evolution, and rupture so that a secondary-creep fit is not mistaken for a full-life model. This skill supports scientific and engineering analysis. It does not replace high-temperature design codes, qualified material databases, inspection evidence, or safety review. Creep is time-dependent inelastic deformation under sustained stress, commonly accelerated by temperature. Whether creep is…

The file

Show its rules and text

We split this skill into the rules it gives an agent

18 rules. About 0 are common, 18 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈18

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Do not apply a steady-state power law as a full-life model when primary strain, tertiary damage, necking, or rupture matters.

    line 72
  2. Distinctive:

    Do not transfer parameters between heat treatments, product forms, weld zones, irradiation states, environments, or alloy batches without …

    line 73
  3. Distinctive:

    Do not fit Arrhenius behavior across a phase or mechanism transition with one activation energy.

    line 74
  4. Distinctive:

    Do not use uniaxial rupture correlations as a multiaxial damage law without validation.

    line 75
  5. Distinctive:

    Do not extrapolate laboratory tests by orders of magnitude without model-form comparison and uncertainty bounds.

    line 76
  6. Distinctive:

    Do not interpret a converged FE time integration as proof that the constitutive or rupture mechanism is valid.

    line 77
  7. Distinctive:

    Test load changes and relaxation histories because constant-load curves may not identify path dependence.

    line 205
  8. Distinctive:

    Check whether fitted activation behavior changes across phase, precipitation, oxidation, or mechanism transitions.

    line 206
  9. Distinctive:

    Do not concatenate curves with different rupture definitions, extensometer conventions, or heat treatments merely to increase sample count.

    line 208
  10. Distinctive:

    Verify the constitutive update with a single integration-point test under constant stress, fixed strain, load change, and temperature …

    line 213
  11. Distinctive:

    Check reactions, energy, local integration status, cutbacks, and state-variable bounds.

    line 217
  12. Distinctive:

    Use FEM multiphysics for boundary conditions, solver tolerances, coupling, and mesh/time-step convergence.

    line 220
  13. Distinctive:

    Use surrogate validation for grouped splits and calibration diagnostics.

    line 236
  14. Distinctive:

    Use a common data contract, objective function, weighting policy, censoring treatment, and validation set.

    line 240
  15. Distinctive:

    Prefer the simplest model that reproduces the observables and paths needed for the decision.

    line 249
  16. Distinctive:

    Use kelvin, explicit stress units, explicit time units, and a dimensionless stress ratio in fitted equations.

    line 281
  17. Distinctive:

    Verify the local constitutive update before running a component model.

    line 285
  18. Distinctive:

    Treat grain size, precipitate state, oxidation, and phase evolution as state variables or domain limits when they change during service.

    line 291

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Creep Mechanics Workflow

Description

Select, calibrate, implement, and validate creep deformation and rupture models for materials and components under sustained thermomechanical loading. The workflow separates rate prediction, transient deformation, stress redistribution, damage evolution, and rupture so that a secondary-creep fit is not mistaken for a full-life model.

This skill supports scientific and engineering analysis. It does not replace high-temperature design codes, qualified material databases, inspection evidence, or safety review.

Domain Context

Creep is time-dependent inelastic deformation under sustained stress, commonly accelerated by temperature. Whether creep is engineering-relevant depends on material class, microstructure, stress, temperature, time, environment, and loading history; no universal temperature or homologous-temperature threshold defines its onset. [EXPERT REVIEW NEEDED]

A conventional uniaxial creep curve may contain:

  • Primary creep: decreasing strain rate as hardening or internal resistance develops.
  • Secondary or minimum-rate region: approximately steady or minimum strain rate over part of the test; a truly constant regime may be absent.
  • Tertiary creep: accelerating strain from damage, necking, cavitation, microstructural evolution, or increasing true stress, ending in rupture.

These stages are curve descriptions rather than unique mechanisms. Diffusion-controlled creep, dislocation glide/climb, grain-boundary sliding, precipitate evolution, recovery, cavitation, oxidation, and phase changes can overlap. Mechanism maps guide hypothesis formation but do not replace alloy- and state-specific evidence.

Homologous temperature uses absolute temperature, normally T/T_m, but the relevant melting or transformation temperature can be ambiguous for multiphase alloys, ceramics, polymers, and evolving materials. [EXPERT REVIEW NEEDED]

Creep analysis has several distinct outputs:

  • deformation or clearance at a service time;
  • stress relaxation under constrained strain;
  • redistributed multiaxial stress and strain fields;
  • onset and evolution of creep damage;
  • time to a specified strain, crack, loss of section, or rupture event;
  • probability or lower-bound life under material and loading scatter.

When to Use This Skill

  • Fit steady or minimum creep rates across stress and temperature.
  • Calibrate primary-secondary-tertiary or internal-variable constitutive models.
  • Simulate creep deformation, relaxation, redistribution, and damage in finite elements.
  • Assess rupture correlations and long-time extrapolation with uncertainty.
  • Compare competing mechanism-based, empirical, and continuum-damage models.
  • Evaluate multiaxial components, weld regions, anisotropy, gradients, or evolving material states.
  • Design validation tests that separate interpolation from service-life extrapolation.

When Not to Use This Skill

  • Do not apply a steady-state power law as a full-life model when primary strain, tertiary damage, necking, or rupture matters.
  • Do not transfer parameters between heat treatments, product forms, weld zones, irradiation states, environments, or alloy batches without evidence.
  • Do not fit Arrhenius behavior across a phase or mechanism transition with one activation energy.
  • Do not use uniaxial rupture correlations as a multiaxial damage law without validation.
  • Do not extrapolate laboratory tests by orders of magnitude without model-form comparison and uncertainty bounds.
  • Do not interpret a converged FE time integration as proof that the constitutive or rupture mechanism is valid.

Core Concepts

Stress and temperature dependence

A common steady-rate representation is

epsilon_dot = A * (sigma / sigma_ref)^n * exp(-Q / (R * T))

where T is absolute temperature, Q and R use consistent energy-per-mole units, and sigma_ref makes the power-law argument dimensionless. The fitted A, , and Q` are effective parameters over the calibration domain. A change in slope with stress or temperature may indicate mechanism transition, threshold stress, microstructure evolution, or data limitations rather than one global exponent.

Diffusion-controlled models often predict lower stress exponents and explicit grain-size dependence, while dislocation-controlled power-law creep can show stronger stress dependence. Exact exponents and activation energies are not mechanism identifiers by themselves. [EXPERT REVIEW NEEDED]

Transient formulations

Time-hardening models express creep rate as a function of stress, temperature, and elapsed time; strain-hardening models use accumulated creep strain as an internal history measure. Their predictions differ under variable load, unloading, and stress redistribution. Internal-variable models can represent hardening, recovery, back stress, damage, precipitate state, or other evolving resistance, but additional parameters increase identifiability demands.

Stress relaxation is a coupled boundary-value problem: under fixed total strain, elastic stress falls as creep strain accumulates. Substituting the initial stress into a constant-stress creep equation is generally inconsistent.

Multiaxial creep and redistribution

Many isotropic implementations drive deviatoric creep with an equivalent stress and assume nearly incompressible creep flow. Hydrostatic stress can still strongly influence cavity growth, grain-boundary damage, and rupture. Choice of von Mises, maximum principal, Huddleston, stress triaxiality, or another rupture/damage measure is model- and material-dependent. [EXPERT REVIEW NEEDED]

Components redistribute load as high-stress regions creep. The life-controlling location can therefore move, and initial elastic peaks may relax while other regions accumulate strain or damage. Weldments and anisotropic products require region- or orientation-specific behavior where evidence supports it.

Damage and rupture

Continuum damage models introduce internal variables that reduce resistance or accelerate strain rate. Damage is model-defined and usually not directly observable; calibrate it against multiple observables such as full creep curves, reduction of area, cavity density, stiffness, and rupture time. Local softening can cause mesh dependence unless the model or failure criterion is regularized or otherwise objective. [EXPERT REVIEW NEEDED]

A Larson-Miller-type parameter is commonly written

P_LM = T * (C + log10(t_r))

but the logarithm base, time unit, fitted constant, temperature unit, stress relation, and calibration window must be stated. C is not a universal constant. [EXPERT REVIEW NEEDED]

A generalized Monkman-Grant relation connects minimum creep rate and rupture time,

(epsilon_dot_min)^m * t_r = C_MG

as an empirical correlation. It does not represent tertiary damage or guarantee transfer across stress, temperature, microstructure, or ductility regimes. [EXPERT REVIEW NEEDED]

Creep-fatigue interaction requires cyclic plasticity, dwell-dependent creep/relaxation, oxidation or environment where relevant, and an interaction rule validated for the intended waveform. Summing independent fatigue and creep fractions is not a universal physical law. [EXPERT REVIEW NEEDED]

Numerical integration and identifiability

Implicit FE integration generally solves local constitutive residuals with Newton iterations and a consistent or numerically suitable tangent. Large time increments can skip rapid primary response, tertiary acceleration, or load redistribution even when global equilibrium converges. Converge deformation, stress, damage, and predicted life with respect to both mesh and time increments.

Creep parameters are often strongly correlated. Sparse tests over a narrow stress-temperature window may identify a curve but not A, , Q`, hardening, and damage parameters separately. Use grouped validation, profile likelihood, bootstrap, Bayesian inference, or comparable uncertainty analysis before long-time extrapolation.

Key Workflows

Workflow 1: Fit steady or minimum creep parameters

Require specimen-level stress, absolute temperature, minimum or steady-rate definition, time unit, strain measure, material state, environment, and test provenance. Fit in log-rate space only after deciding whether the residual model is appropriate; log transformation changes the error assumptions.

import numpy as np
from scipy.optimize import least_squares

R_GAS = 8.314462618  # J mol^-1 K^-1


def fit_norton_arrhenius(
    stress_pa, temperature_K, rate_per_s, sigma_ref_pa, initial, bounds
):
    """Fit log(A), n, and Q[J/mol] to positive finite creep-rate data."""
    stress = np.asarray(stress_pa, dtype=float)
    temp = np.asarray(temperature_K, dtype=float)
    rate = np.asarray(rate_per_s, dtype=float)
    if stress.shape != temp.shape or stress.shape != rate.shape or stress.ndim != 1:
        raise ValueError("Stress, temperature, and rate must be equal-length vectors")
    if len(rate) < 4 or not np.isfinite(np.column_stack([stress, temp, rate])).all():
        raise ValueError("Provide at least four finite observations")
    if (
        (stress <= 0).any()
        or (temp <= 0).any()
        or (rate <= 0).any()
        or not np.isfinite(sigma_ref_pa)
        or sigma_ref_pa <= 0
    ):
        raise ValueError("Stress, absolute temperature, rate, and reference stress must be positive")

    initial = np.asarray(initial, dtype=float)
    lower, upper = (np.asarray(bound, dtype=float) for bound in bounds)
    if initial.shape != (3,) or lower.shape != (3,) or upper.shape != (3,):
        raise ValueError("initial and both bounds must contain log(A), n, and Q")
    if not np.isfinite(np.concatenate([initial, lower, upper])).all():
        raise ValueError("Initial values and bounds must be finite")
    if not ((lower < initial).all() and (initial < upper).all()):
        raise ValueError("Initial values must lie strictly inside ordered bounds")

    def residual(theta):
        log_A, n, Q_J_per_mol = theta
        predicted_log_rate = (
            log_A + n * np.log(stress / sigma_ref_pa) - Q_J_per_mol / (R_GAS * temp)
        )
        return predicted_log_rate - np.log(rate)

    result = least_squares(residual, x0=initial, bounds=(lower, upper))
    if not result.success:
        raise RuntimeError(result.message)
    log_A, n, Q = result.x
    A = np.exp(log_A)
    if not np.isfinite([A, n, Q]).all() or A <= 0 or n <= 0 or Q <= 0:
        raise RuntimeError("Fit returned nonphysical or nonfinite parameters")
    return {
        "A_per_s": float(A),
        "n": float(n),
        "Q_J_per_mol": float(Q),
        "result": result,
    }

Set bounds from defensible material knowledge and the experimental domain; no universal bounds apply. [EXPERT REVIEW NEEDED] Inspect residuals against stress, temperature, time, batch, and specimen. Bootstrap by specimen or batch, not by individual rows from the same test. Enforce no mechanism interpretation from n or Q without independent evidence.

Workflow 2: Calibrate temperature-dependent transient creep

  1. Convert every temperature to kelvin and every time to one declared unit.
  2. Align zero strain/time, machine compliance, thermal strain, and load-ramp treatment.
  3. Fit full curves or carefully selected features across multiple stresses and temperatures.
  4. Compare time-hardening, strain-hardening, theta-projection, internal-variable, or mechanistic candidates using held-out histories.
  5. Test load changes and relaxation histories because constant-load curves may not identify path dependence.
  6. Check whether fitted activation behavior changes across phase, precipitation, oxidation, or mechanism transitions.

Do not concatenate curves with different rupture definitions, extensometer conventions, or heat treatments merely to increase sample count.

Workflow 3: Simulate creep deformation and stress relaxation in FEM

  1. Establish the thermal and mechanical initial state, including residual stress and preload.
  2. Verify the constitutive update with a single integration-point test under constant stress, fixed strain, load change, and temperature change.
  3. Compare FE uniaxial output against the calibration curves using the same true/engineering stress and strain conventions.
  4. Converge time increments during primary response, stress redistribution, and tertiary acceleration.
  5. Converge the mesh on displacement, stress, accumulated creep strain, damage, and life-controlling location.
  6. Check reactions, energy, local integration status, cutbacks, and state-variable bounds.
  7. Validate component deformation or relaxation against independent measurements where available.

Use FEM multiphysics for boundary conditions, solver tolerances, coupling, and mesh/time-step convergence.

Workflow 4: Assess creep rupture or time to a limit

Define the event: specified creep strain, onset of tertiary acceleration, crack indication, loss of load, or complete rupture. Fit Larson-Miller, Monkman-Grant, continuum-damage, or another model only over a documented stress-temperature-time window. Retain runouts and interrupted tests with correct censoring. Compare predictions with direct long-duration tests withheld from calibration, and report both interpolation and extrapolation distances.

For component life, integrate the constitutive/damage history under redistributed stress rather than applying a uniaxial rupture time to the initial peak stress. Apply any design-code reduction factors and allowable definitions exactly as specified by the governing code.

Workflow 5: Analyze multiaxial component creep

Partition base material, weld metal, heat-affected zones, coatings, and anisotropic regions when evidence supports different behavior. Simulate sustained loads and temperature gradients through the service sequence. Track equivalent and principal stresses, triaxiality, creep strain, damage, constraint, and location migration. Compare candidate multiaxial rupture measures against notched-bar, tube, cross-weld, or other relevant validation tests before relying on component predictions.

Workflow 6: Quantify parameter uncertainty and extrapolation risk

Create grouped calibration and validation partitions by heat, specimen, stress level, and temperature. Sample parameter uncertainty and model-form alternatives, then propagate them through service histories. Plot prediction intervals against time and explicitly mark the convex hull of calibration conditions. Perform leave-one-stress and leave-one-temperature-out tests; a random row split is not a meaningful extrapolation test for correlated creep curves.

Use surrogate validation for grouped splits and calibration diagnostics. Long-life tails can be dominated by model discrepancy rather than fitted-parameter variance; report both where possible. [EXPERT REVIEW NEEDED]

Workflow 7: Compare competing creep models

Use a common data contract, objective function, weighting policy, censoring treatment, and validation set. Compare at minimum:

  • minimum-rate prediction across stress and temperature;
  • full strain-time curves and relaxation histories;
  • rupture or limit-event time;
  • parameter stability under data removal;
  • extrapolation behavior and physical admissibility;
  • FE robustness, time-step sensitivity, and computational cost.

Prefer the simplest model that reproduces the observables and paths needed for the decision. A lower calibration error does not justify an unidentifiable or physically implausible model.

Common Failure Modes

FailureSymptomLikely CauseDiagnosticFix
Power law extrapolated across unsupported stressesLarge life/rate errors at stress extremesMechanism or threshold behavior changedPlot local slopes and leave-one-stress-out predictionsRestrict domain or use a supported transition model
Inappropriate temperature transferArrhenius fit fails at new temperaturePhase, diffusion path, or microstructure changedResiduals and microscopy versus temperaturePartition regimes or adopt state-aware behavior
Secondary creep treated as full lifetimeDeformation fits but rupture is missedPrimary and tertiary stages omittedCompare full curves and rupture timesAdd transient and damage/rupture modeling
Tertiary damage ignoredNonconservative remaining lifeCavitation, necking, or softening absentInspect accelerating rate, area loss, and damage evidenceCalibrate a validated full-life model
Sparse long-duration data overfitUnstable extrapolationToo many correlated parametersProfiles, bootstrap, and held-out long testsReduce model complexity and widen uncertainty
Activation energy units wrongImplausible rate changeskJ/mol mixed with J/mol or wrong gas constantDimensional audit and manual exponent checkConvert Q and R consistently
Celsius used in Arrhenius termCatastrophic temperature errorAbsolute temperature not usedAssert kelvin and inspect value rangeConvert to kelvin at ingestion
Time-unit mismatchRate/life error by factors of 3600 or moreHours and seconds mixedRecompute one observation manuallyCanonicalize time and record source units
Stress units hidden in coefficientParameters cannot transfer between scriptsDimensional power law used without reference stressInspect equation and units of AUse dimensionless stress ratio or explicit units
Engineering and true quantities mixedArtificial tertiary acceleration or rate biasArea change and strain definitions inconsistentReprocess raw force/extension consistentlyDeclare and transform measures appropriately
Poor parameter identifiabilityLarge parameter changes with similar curvesNarrow stress-temperature-path coverageCorrelation matrix, profiles, and data-removal testsAdd discriminating histories or simplify model
Minimum rate selected inconsistentlyScatter tied to analyst choiceSmoothing/window policy variesRepeat automated and blinded extractionDefine and version the extraction method
Stress redistribution neglectedWrong hot spot and lifeInitial elastic field treated as permanentCompare transient FE stress mapsIntegrate creep under evolving stress
Constant-stress law used for relaxationStress decay is inconsistentFixed total strain not enforcedSingle-element fixed-strain benchmarkSolve coupled elastic-creep evolution
Excessive FE time incrementsMissed primary or tertiary responseAdaptive stepping governed only by equilibriumTime-step convergence of state variablesLimit increments by constitutive evolution
Constitutive integration errorsNegative damage, oscillations, or cutbacksLocal Newton/tangent/state update is faultyMaterial-point tests and residual historiesCorrect integration, tangent, and substepping
Damage localization is mesh dependentRupture time changes with refinementLocal softening lacks objectivityMesh study of dissipation and lifeRegularize or use an objective failure measure
Uniaxial rupture law used multiaxiallyWrong failure locationHydrostatic/maximum stress effects omittedCompare notched or multiaxial validation testsCalibrate a suitable multiaxial measure
Parameters transferred across alloy statesStrong batch or heat-treatment biasMicrostructure differsStratify residuals by material stateCalibrate state-specific or hierarchical models
Rupture correlation used outside its windowImplausible long-time lifeLarson-Miller or other relation extrapolatedMark calibration hull and withheld testsRestrict use and propagate model discrepancy
Creep-fatigue fractions treated as universalDwell/waveform ranking failsInteraction and oxidation omittedCompare representative dwell-cycle testsUse a validated interaction framework
Fit quality confused with mechanism validityAccurate interpolation, wrong response to path changesEmpirical flexibility hides wrong physicsValidate load changes, temperatures, and microstructureSeparate predictive validation from mechanism claims

Best Practices

  1. Define the requested observable and terminal event before selecting a creep equation.
  2. Preserve raw force, extension, temperature, time, atmosphere, and specimen geometry.
  3. Use kelvin, explicit stress units, explicit time units, and a dimensionless stress ratio in fitted equations.
  4. Retain full creep curves instead of storing only minimum rate and rupture time.
  5. Partition specimens by heat, product form, weld region, orientation, and environment.
  6. Calibrate constant-load, load-change, and relaxation behavior when service paths require them.
  7. Verify the local constitutive update before running a component model.
  8. Converge both mesh and time increment on the final deformation and life decisions.
  9. Track evolving true stress and geometry when deformation is no longer small.
  10. Separate rate, deformation, damage, and rupture validation.
  11. Quantify parameter correlation, censored observations, and model discrepancy.
  12. Compare multiple extrapolation models and preserve direct long-duration validation data.
  13. Treat grain size, precipitate state, oxidation, and phase evolution as state variables or domain limits when they change during service. [EXPERT REVIEW NEEDED]
  14. Preserve software version, material subroutine, state-variable definitions, solver tolerances, and restart behavior.
  15. Require engineering review when results support allowable stress, remaining life, inspection, or safety decisions.

Integration with Other Skills

  • fem-multiphysics: Implement transient thermal-mechanical fields, nonlinear solvers, and verified boundary conditions.
  • crystal-plasticity-workflow: Represent anisotropic slip, grain interactions, and microstructure-sensitive creep where data support the added resolution.
  • fatigue-analysis-workflow: Couple dwell, relaxation, and cyclic response in creep-fatigue assessments without using universal damage fractions.
  • phase-field-workflow: Model coupled microstructure or damage evolution only with calibrated thermodynamics, kinetics, and length scales.
  • reproducibility-workflow: Preserve test provenance, fitting code, material models, and FE state definitions.
  • scientific-visualization: Plot full curves, rate minima, residuals, redistribution, damage, censoring, and extrapolation domains.
  • small-data-ml: Fit sparse creep datasets with grouped validation and physically informed constraints.
  • surrogate-validation: Test uncertainty calibration and distribution shift across stress, temperature, and material state.

Key References

  1. Norton, F. H. (1929). The Creep of Steel at High Temperatures. McGraw-Hill. https://books.google.com/books?id=vswxAAAAMAAJ
  2. Frost, H. J., & Ashby, M. F. (1982). Deformation-Mechanism Maps: The Plasticity and Creep of Metals and Ceramics. Pergamon Press. https://engineering.dartmouth.edu/defmech/
  3. Riedel, H. (1987). Fracture at High Temperatures. Springer. https://doi.org/10.1007/978-3-642-82961-1
  4. Larson, F. R., & Miller, J. (1952). A time-temperature relationship for rupture and creep stresses. Transactions of the ASME, 74(5), 765-771. https://doi.org/10.1115/1.4015909
  5. Monkman, F. C., & Grant, N. J. (1956). An empirical relationship between rupture life and minimum creep rate in creep-rupture tests. Proceedings of ASTM, 56, 593-620. No DOI located in the bibliographic sources used for verification.
  6. Hayhurst, D. R. (1972). Creep rupture under multi-axial states of stress. Journal of the Mechanics and Physics of Solids, 20(6), 381-382. https://doi.org/10.1016/0022-5096(72)90015-4
  7. Dyson, B. (2000). Use of CDM in materials modeling and component creep life prediction. Journal of Pressure Vessel Technology, 122(3), 281-296. https://doi.org/10.1115/1.556185
  8. Rabotnov, Y. N. (1969). Creep Problems in Structural Members. North-Holland. No DOI located in the bibliographic sources used for verification.
  9. ASTM International. (2024). ASTM E139-24: Standard Test Methods for Conducting Creep, Creep-Rupture, and Stress-Rupture Tests of Metallic Materials. https://doi.org/10.1520/E0139-24
  10. International Organization for Standardization. (2023). ISO 204:2023 Metallic materials - Uniaxial creep testing in tension - Method of test (4th ed.). https://www.iso.org/standard/82162.html

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.