The START project can optionally clone several Active Inference and
computational neuroscience repositories into src/_clones/ for local
inspection and curriculum development. Cloned content is reference material
only: it is not automatically treated as verified evidence, imported into
generated curricula, or included in the release bundle.
src/_clones/cognitivesrc/_clones/gnn8f293d9ba810e6eda5c09842d9c7ed9ffa5f92c1 (v5.0.3) — the registered revision recorded in src/repos/cloning.pystart-gnn runs the pipeline in this checkout and reports the artifacts it produced.src/_clones/cerebrumsrc/_clones/rxinfersrc/_clones/activeinferencetextbook (educational focus)src/_clones/pymdptextbook (educational focus)src/_clones/lean_nichesrc/_clones/templatesrc/_clones/axiomSteps
git pullLinks
uv run start-clone (the module is also importable as a Python API)uv run start-gnn (see Running the GNN Pipeline)Use the Repository Manager from the main runner for a guided flow:
./run.sh
# Choose: "Repository Manager"
Use the integrated clone utility for consistent repository management:
# Core knowledge graph
uv run start-clone --url https://github.com/ActiveInferenceInstitute/cognitive --dest src/_clones/cognitive --shallow
# Generalized Notation Notation (GNN)
uv run start-clone --url https://github.com/ActiveInferenceInstitute/Generalized_Notation_Notation --dest src/_clones/gnn --shallow
# CEREBRUM
uv run start-clone --url https://github.com/ActiveInferenceInstitute/CEREBRUM --dest src/_clones/cerebrum --shallow
# Bayesian inference examples (RxInfer.jl)
uv run start-clone --url https://github.com/docxology/RxInferExamples.jl --dest src/_clones/rxinfer --shallow
# Julia Active Inference (textbook branch)
uv run start-clone --url https://github.com/docxology/ActiveInference.jl --dest src/_clones/activeinference --branch textbook --shallow
# Python Active Inference (textbook branch)
uv run start-clone --url https://github.com/docxology/pymdp --dest src/_clones/pymdp --branch textbook --shallow
# Lean theorem proving (LeanNiche)
uv run start-clone --url https://github.com/docxology/lean_niche --dest src/_clones/lean_niche --shallow
# Research manuscript template (Thin Orchestrator)
uv run start-clone --url https://github.com/docxology/template --dest src/_clones/template --shallow
# VERSES AXIOM
uv run start-clone --url https://github.com/VersesTech/axiom --dest src/_clones/axiom --shallow
# Clone with specific options
git clone --shallow-since="2023-01-01" --branch textbook https://github.com/docxology/ActiveInference.jl src/_clones/activeinference
# Update existing clones
cd src/_clones/cognitive && git pull origin main
cd src/_clones/pymdp && git pull origin textbook
The GNN checkout contains its own 25-step pipeline. start-gnn runs a bounded
selection (default steps 3, 5, 11, 12: parse, type-check, render, execute) and
reports what it produced: per-step statuses, artifact counts, artifact hashes,
and the models the pipeline recorded as consumed.
# Default: steps 3,5,11,12 over input/gnn_files/basics
uv run start-gnn
# Full exemplar corpus, with a machine-readable report
uv run start-gnn --target input/gnn_files --json
# Print the command without executing it
uv run start-gnn --dry-run
# Clone the registered repository first when the checkout is missing
uv run start-gnn --clone
# Fail closed when the checkout is not the registered revision
uv run start-gnn --require-pin
Behaviour
.venv when present, then
uv run --frozen --no-sync python (no environment mutation), then the
interpreter running START. Override with --python-executable.<checkout>/output;
--output-dir <path> keeps a run outside the checkout.start_gnn_summary.json, start_gnn_stdout.log and (when
non-empty) start_gnn_stderr.log beside the pipeline artifacts.at_revision, ahead_of_revision, diverged, or unknown when a
shallow clone cannot resolve the revision object).0 pipeline ran, 1 pipeline failed, 2 run refused.src/_clones/
├── cognitive/ # Knowledge graph and concept definitions
│ ├── content/ # Structured Active Inference content
│ └── ontologies/ # Formal concept relationships
├── gnn/ # Generalized Notation Notation
├── cerebrum/ # CEREBRUM
├── rxinfer/ # Bayesian inference examples (RxInfer.jl)
│ ├── notebooks/ # Jupyter notebooks with examples
│ └── scripts/ # Standalone example scripts
├── activeinference/ # Julia Active Inference implementation
│ ├── docs/ # Educational documentation
│ ├── examples/ # Code examples and tutorials
│ └── src/ # Core implementation
├── pymdp/ # Python Active Inference
│ ├── notebooks/ # Educational notebooks
│ ├── examples/ # Example scripts and demonstrations
│ └── pymdp/ # Core Python package
├── lean_niche/ # Lean theorem proving & verification environment
│ ├── src/ # Lean and Python sources
│ ├── docs/ # Project docs and verification workflows
│ └── examples/ # Example proofs and verification scripts
├── template/ # Thin orchestrator research manuscript template
└── axiom/ # VERSES AXIOM knowledge operating system
# Path exists and is a git repo
test -d src/_clones/cognitive/.git && echo OK
# Show remote and branch
git -C src/_clones/cognitive remote -v
git -C src/_clones/cognitive branch --show-current
git -C src/_clones/cognitive fetch --prune
git -C src/_clones/cognitive pull --ff-only origin main
git -C src/_clones/pymdp fetch --prune
git -C src/_clones/pymdp pull --ff-only origin textbook
rm -rf src/_clones/activeinference
uv run start-clone --url https://github.com/docxology/ActiveInference.jl --dest src/_clones/activeinference --branch textbook --shallow