Disclaimer & Assessment Methodology

This score was produced via LLM-assisted assessment of the repository structure and the supplied evidence bundle. These findings should be treated as a directional pilot signal rather than a verified rating. For more details on our pilot objectives and the development path toward rules-based human review, see the Development Roadmap. Raw files are accessible at the assessment report source.

Evidence Chain Assessment

PyPSA-Eur Assessment

Assessed Claim

"An open sector-coupled optimisation model of the European energy system."

PyPSA-Eur shows strong Open Traceability overall for an open European energy-system optimisation model: its public repository, open-source licensing statements, dependency/environment files, Snakemake-based workflow description, CI/integration tests, validation workflows, community contribution processes, documentation, citation metadata, Zenodo links, and explicit limitations make much of the evidence and method chain externally inspectable. The strongest dimensions are open-source software and public communication, while input data and reproducibility are also well supported by named data sources, versioning infrastructure, and automated workflows. The main traceability gap is exact end-to-end linkage for individual environmental claims or published results: the supplied evidence does not include a complete provenance record tying a specific claim to exact primary data versions, code commit, model configuration, solver settings, workflow logs, output files, review records, and final publication artifact.

77 Traceability Score
Substantial Traceability
Project Source Source Repository
Framework Version v1.0.0
Assessment Date 2026-06-12
Human Review Pending

Claim Traceability Pathway

Inspection trail mapping the connections from raw inputs to the published claim.

Stage 01 Input Data
Score: 74
Stage 02 Models & Software
Score: 88
Stage 03 Execution Run
Score: 76
Stage 04 Community Review
Score: 77
Stage 05 Publication
Score: 83
Target Claim Stated Claim
Claim
Strong inspectability (≥80)
Moderate inspectability (55-79)
Weak inspectability (<55)
Open Linkage Connection

Traceability Health Summary

Scan diagnosis of where the evidence chain stands and what requires attention first.

Strongest Area

Open-Source Models, Methods, and Software

Inspectability: 88/100

This stage provides solid evidence, making calculations and documentation highly inspectable and reusable for external actors.

Traceability Gap

Open Linkage

Inspectability: 64/100

This is the weakest segment of the evidence chain. Gaps in records or lack of linkage break the evidence chain here.

Missing Artifacts
  • Unique dataset DOIs & licenses
  • Dynamic provenance graph mapping
First Priority Action

Strengthen Open Linkage

Create a versioned provenance graph mapping every step of the computation from source data to claim.

64 Traceability Score
Framework Metric

Evidence Linkage Summary

Open Linkage is not a standalone category; it is the connectivity of the evidence chain. Below is the connection status of the connection links between each step in this project's journey.

Linkage Summary

The project has moderate-to-strong linkage evidence across the chain: the README connects the model claim to named input datasets, Snakemake workflow construction, documentation, limitations, Zenodo records, releases, and papers; CITATION.cff identifies a specific version and repository; the data-versioning validation layer implies structured links among datasets, versions, sources, tags, dates, and URLs; and CI workflows connect code revisions to tests, validation, release preparation, and container images tagged by commit SHA. However, the evidence does not show an explicit, externally inspectable provenance graph from a specific environmental claim or publication result to exact data versions, code commit, model configuration, solver environment, workflow run logs, outputs, review records, and publication artifacts. The linkage is therefore substantially present at project-infrastructure level but incomplete at exact claim/run level. It is not lower because several artifacts deliberately support versioning and cross-referencing; it is not higher because exact end-to-end claim-level traceability is not demonstrated in the supplied bundle.

Detailed Dimension Inspection

Granular findings for each processing step in the Open Traceability framework.

Stage 01

Open Input Data and Measurement Evidence

Score: 74/100

Summary: PyPSA-Eur provides strong public evidence that major model inputs are identifiable and based on open or externally accessible sources: the README lists the power grid model, power plant database, ENTSO-E demand, ERA5/SARAH-3 renewable time series via atlite, land eligibility analysis, and Eurostat/JRC-IDEES energy balances. The repository also includes dataset documentation for at least one input layer, entsoegridkit, with fields, provenance, processing caveats, and warnings about inaccuracies. There is evidence of a data-versioning layer through tests validating data/versions.csv, including source categories, archive URLs, tags such as latest/supported/deprecated, and consistency checks. The score is not higher because the evidence bundle does not include the actual data/versions.csv, complete source manifest, licences and access conditions for every input, quantified uncertainty/error estimates for all primary datasets, or full transformation documentation from primary to model-ready data. It is not lower because the supplied evidence clearly shows a systematic, documented open-data approach and explicit caveats for input quality.

Evidence Found

PyPSA-Eur provides strong public evidence that major model inputs are identifiable and based on open or externally accessible sources: the README lists the power grid model, power plant database, ENTSO-E demand, ERA5/SARAH-3 renewable time series via atlite, land eligibility analysis, and Eurostat/JRC-IDEES energy balances. The repository also includes dataset documentation for at least one input layer, entsoegridkit, with fields, provenance, processing caveats, and warnings about inaccuracies. There is evidence of a data-versioning layer through tests validating data/versions.csv, including source categories, archive URLs, tags such as latest/supported/deprecated, and consistency checks. The score is not higher because the evidence bundle does not include the actual data/versions.csv, complete source manifest, licences and access conditions for every input, quantified uncertainty/error estimates for all primary datasets, or full transformation documentation from primary to model-ready data. It is not lower because the supplied evidence clearly shows a systematic, documented open-data approach and explicit caveats for input quality.

Gaps Identified

No significant gaps identified. This stage is fully documented and verifiable.

Stage 02

Open-Source Models, Methods, and Software

Score: 88/100

Summary: The project has excellent public software traceability: it is hosted in a public GitHub repository, describes itself as fully open source, provides citation metadata identifying version v2026.02.0 and MIT licensing, includes SPDX licence headers in supplied source/test files, lists a full Conda environment with modelling and geospatial dependencies, and provides a Dockerfile for a development environment. The repository uses version control, release badges, dependency-management tooling, and open-source frameworks such as PyPSA, Snakemake, atlite, and powerplantmatching. The score is not higher because GitHub repository metadata reports no repository-level licence, the evidence bundle does not include the root LICENSE file or pixi.lock, and environment.yaml uses version ranges rather than a fully locked environment. It is not lower because the available evidence is explicit, reusable, and strongly supports inspection and operation of the model software.

Evidence Found

The project has excellent public software traceability: it is hosted in a public GitHub repository, describes itself as fully open source, provides citation metadata identifying version v2026.02.0 and MIT licensing, includes SPDX licence headers in supplied source/test files, lists a full Conda environment with modelling and geospatial dependencies, and provides a Dockerfile for a development environment. The repository uses version control, release badges, dependency-management tooling, and open-source frameworks such as PyPSA, Snakemake, atlite, and powerplantmatching. The score is not higher because GitHub repository metadata reports no repository-level licence, the evidence bundle does not include the root LICENSE file or pixi.lock, and environment.yaml uses version ranges rather than a fully locked environment. It is not lower because the available evidence is explicit, reusable, and strongly supports inspection and operation of the model software.

Gaps Identified

No significant gaps identified. This stage is fully documented and verifiable.

Stage 03

Open Execution and Reproducibility

Score: 76/100

Summary: The project has strong execution-traceability infrastructure: the README says the model is built from open data using a Snakemake workflow; the environment file documents dependencies; CI runs unit tests, integration tests, pylint checks, and Snakemake test workflows; validation workflows run self-hosted model validation and generate result comments; release workflows update documentation DAGs; and container-image workflows tag development environments by commit SHA. These artifacts make the computational pathway substantially inspectable and partly re-executable. The score is not higher because the evidence bundle does not include the Snakefile/rules, exact scenario configuration for a published environmental claim, full logs, result artifacts, solver settings, random seeds if any, or a specific run record linking inputs to outputs. It is not lower because there is clear evidence of automated tests, integration workflows, environment specification, and reproducibility-oriented workflow tooling.

Evidence Found

The project has strong execution-traceability infrastructure: the README says the model is built from open data using a Snakemake workflow; the environment file documents dependencies; CI runs unit tests, integration tests, pylint checks, and Snakemake test workflows; validation workflows run self-hosted model validation and generate result comments; release workflows update documentation DAGs; and container-image workflows tag development environments by commit SHA. These artifacts make the computational pathway substantially inspectable and partly re-executable. The score is not higher because the evidence bundle does not include the Snakefile/rules, exact scenario configuration for a published environmental claim, full logs, result artifacts, solver settings, random seeds if any, or a specific run record linking inputs to outputs. It is not lower because there is clear evidence of automated tests, integration workflows, environment specification, and reproducibility-oriented workflow tooling.

Gaps Identified

No significant gaps identified. This stage is fully documented and verifiable.

Stage 04

Open Community and Review

Score: 77/100

Summary: The project provides strong evidence of open community and review mechanisms: contri

Evidence Found

The project provides strong evidence of open community and review mechanisms: contri

Gaps Identified

buting documentation welcomes bug reports, ideas, issues, pull requests, help-wanted work, and Discord discussion; it explicitly says all code contri

Stage 05

Open Publications and Communication

Score: 83/100

Summary: PyPSA-Eur has strong publication and communication traceability. The README is openly accessible, clearly states the model’s purpose and scope, lists the evidence base and major data sources, includes warnings and limitations, links to documentation, release badges, Zenodo DOIs, and example peer-reviewed or scholarly outputs. CITATION.cff provides a citable project record with version, authors, repository, and licence. Documentation files in the bundle use CC-BY-4.0 SPDX headers, supporting reuse of explanatory material. The score is not higher because the bundle does not include the full documentation, complete dashboard/update-cycle documentation if any, full release notes, paper licences, or a complete mapping from each publication claim to exact artifacts. It is not lower because the main public communication artifacts are open, well labelled, citable, and explicitly point users to limitations and supporting resources.

Evidence Found

PyPSA-Eur has strong publication and communication traceability. The README is openly accessible, clearly states the model’s purpose and scope, lists the evidence base and major data sources, includes warnings and limitations, links to documentation, release badges, Zenodo DOIs, and example peer-reviewed or scholarly outputs. CITATION.cff provides a citable project record with version, authors, repository, and licence. Documentation files in the bundle use CC-BY-4.0 SPDX headers, supporting reuse of explanatory material. The score is not higher because the bundle does not include the full documentation, complete dashboard/update-cycle documentation if any, full release notes, paper licences, or a complete mapping from each publication claim to exact artifacts. It is not lower because the main public communication artifacts are open, well labelled, citable, and explicitly point users to limitations and supporting resources.

Gaps Identified

No significant gaps identified. This stage is fully documented and verifiable.

Stage 06

Open Linkage

Score: 64/100

Summary: The project has moderate-to-strong linkage evidence across the chain: the README connects the model claim to named input datasets, Snakemake workflow construction, documentation, limitations, Zenodo records, releases, and papers; CITATION.cff identifies a specific version and repository; the data-versioning validation layer implies structured links among datasets, versions, sources, tags, dates, and URLs; and CI workflows connect code revisions to tests, validation, release preparation, and container images tagged by commit SHA.

Evidence Found

The project has moderate-to-strong linkage evidence across the chain: the README connects the model claim to named input datasets, Snakemake workflow construction, documentation, limitations, Zenodo records, releases, and papers; CITATION.cff identifies a specific version and repository; the data-versioning validation layer implies structured links among datasets, versions, sources, tags, dates, and URLs; and CI workflows connect code revisions to tests, validation, release preparation, and container images tagged by commit SHA.

Gaps Identified

However, the evidence does not show an explicit, externally inspectable provenance graph from a specific environmental claim or publication result to exact data versions, code commit, model configuration, solver environment, workflow run logs, outputs, review records, and publication artifacts. The linkage is therefore substantially present at project-infrastructure level but incomplete at exact claim/run level. It is not lower because several artifacts deliberately support versioning and cross-referencing; it is not higher because exact end-to-end claim-level traceability is not demonstrated in the supplied bundle.

Critical Traceability Gaps

Inspection details of where the evidence chain breaks down and what that implies for public trust.

Identified GapImpact on TraceabilitySuggested Remediation
Dataset not linked or versionedHigh ‐ Blocks inspection of inputsRegister raw datasets in Zenodo and cite unique DOIs in config files.

Remediation & Improvements Roadmap

Actionable roadmap prioritized by implementation complexity.

Phase 1

Quick Wins

Low effort, high immediate return.

  • Add CITATION.cff file

    Enables clear publication-to-repository citation schema.

  • Commit top-level LICENSE

    Resolves legal terms of software and dataset inspectability.

  • Link communication channels

    Add status/discord badges to repository README files.

Phase 2

Medium Effort

Standard engineering practices.

  • Specify configurations in config files

    Provide explicit parameters and dataset configurations in repository.

  • Lock computational environments

    Lock dependencies using Conda-lock or Docker image digests.

  • Archive automated workflow logs

    Publish CI test summaries and solver objective check values.

Phase 3

Strategic Upgrades

Highest traceability return.

  • Publish workflow runs on Zenodo

    Create persistent, immutable archives of the full execution outputs.

  • Build claims-to-evidence provenance graphs

    Establish a machine-readable map of the end-to-end evidence trail.

Framework Dimension Definitions

  • Open Input Data and Measurement Evidence: Identifiable, documented, attributable, versioned raw measurement evidence inputs.
  • Open-Source Models, Methods, and Software: Inspectability of Analytical logic via code repositories, configurations, and license files.
  • Open Execution and Reproducibility: Computational environments, workflows, run logs, and outputs make rerun processes inspectable.
  • Open Community and Review: Scrutiny, peer review, issue tracking, and responses to replication challenges are visible.
  • Open Publications and Communication: Resulting reports, papers, and dashboards are openly accessible and reference source materials.
  • Open Linkage: The versioned, explicit connectivity linking data, models, execution, review, and publication steps.

Human Review Notes

Human review of this assessment report is pending. An independent reviewer must inspect the repository references to verify the findings.