Command Reference¶

Use this page when you know the task and want the public command family quickly, without reading the full technical manual.

Read before: Choose Your Path if you are not sure which command family you need.

Need the shortest practical list? Use the Command Cheatsheet.

Read next: Technical Reference for deeper integration details.

Beginner-Friendly Entry Points¶

`iints guide`¶

Use this when you are not sure where to start.

`iints start`¶

Print a goal-based first-run plan, or run the safe starter action.

Common forms:

iints start
iints start --goal project --run
iints start --goal edge
iints start --goal data

`iints onboard`¶

Show the one recommended path from machine check to first study bundle.

iints onboard
iints onboard --run-safe-steps

The safe mode runs doctor, a full demo, demo-data import, and a realism check, then prints the two study commands you can run once you choose an algorithm.

`iints demo`¶

The main live-demo button. It exports showable code, prints the presenter story, runs the demo in audience-safe stage mode, writes a technical log off-screen, and produces the cue card, artifact map, presenter guide, poster, and rerun script. By default it also builds evidence_bundle/ with a public README, model card, run index, and copied proof artifacts.

Common forms:

iints demo
iints demo doctor
iints demo eucys
iints demo booth
iints demo --audience jury --output-dir results/live_demo
iints demo --audience clinical
iints demo --dry-run          # rehearsal/preflight only, not the live call
iints demo --no-evidence      # fast rehearsal without the public proof bundle
iints demo --technical        # show raw execution detail for debugging
iints demo --simulation-only --quick
iints demo --simulation-only --full

Story presets:

iints demo doctor starts with a clinical safety discussion: virtual patient, normal day, meal stress, risky context, and supervisor decision.
iints demo eucys frames the run as one experiment with a research question, hypothesis, three runs, and evidence bundle.
iints demo booth frames the run as a public digital-patient story: scenario changes, algorithm suggestion, safety check, and visual proof.

`iints demo-live`¶

Explicit alias for the same live presentation engine used by iints demo.

It exports showable Python code, prints an audience-aware opening talk track, runs the live demo, writes PRESENTER_GUIDE.md, DEMO_STORY.md, DEMO_CUE_CARD.md, DEMO_ARTIFACTS.md, and RUN_LIVE_DEMO.sh, then lists the poster plus proof artifacts to open next. Non-SDK story modes keep code as the proof layer instead of the first thing shown.

Common forms:

iints demo-live
iints demo-live --output-dir results/live_demo
iints demo-live --no-run
iints demo-live --prepare-ai
iints demo-live --audience clinical
iints demo-live --audience engineering
iints demo-live --story doctor
iints demo-live --story eucys
iints demo-live --story booth

`iints quickstart`¶

Create a ready-to-run project folder.

iints quickstart --project-name iints_quickstart

The generated project is self-contained: it includes patients/stable_patient.yaml, a scenario file, and an editable starter algorithm so you can run locally without depending on packaged patient assets.

`iints run --wizard`¶

Interactive custom run builder.

`iints update`¶

Update the current Python environment to the newest SDK release.

Common forms:

iints update
iints update --dry-run
iints update --repair --force-reinstall --yes
iints update --no-cache-dir --yes
iints update --source github --yes
iints update --extras full,mdmp,research,edge

The command prints the exact python -m pip install -U ... invocation before it changes anything. By default it uses --source auto: PyPI first, then GitHub main as a fallback if the release has not propagated yet. Use --dry-run during a live demo setup check.

`iints delete`¶

Remove IINTS from the current machine/environment with a visible deletion plan.

Common forms:

iints delete --dry-run
iints delete --yes
iints delete --everything --dry-run
iints delete --everything --yes
iints delete --source-checkout --yes
iints delete --local-outputs --yes
iints delete --no-packages --path results/old_iints_run --yes

Default behavior removes the active Python SDK packages plus user-level IINTS config, plugin, and cache folders. --everything also includes known generated output folders in the current directory and a detected local IINTS-SDK source checkout. It still does not guess private datasets, external-drive research archives, or unrelated virtual environments.

Core Simulation Commands¶

`iints run`¶

Run one simulation.

Examples:

iints run --preset baseline_t1d
iints run --algo algorithms/example_algorithm.py --scenario scenarios/example_scenario.json
iints run --dry-run --preset baseline_t1d

`iints run-full`¶

One-line run with full output bundle.

`iints run-parallel`¶

Run a matrix of scenarios in parallel.

`iints benchmark`¶

Compare algorithms across standard workloads.

Extension Commands¶

`iints plugin install`¶

Install a local algorithm plugin without editing SDK source code.

iints plugin install algorithms/my_algo.py
iints algorithms list

The SDK copies the file into the local plugin home and records it in ~/.iints/plugins/registry.json. For tests or portable environments, set IINTS_PLUGIN_HOME to another folder.

`iints plugin register`¶

Register extension files by kind.

iints plugin register algo algorithms/my_algo.py
iints plugin register patient-model patient_models/my_model.py --name "My Model"
iints plugin register data-source data_sources/my_importer.py
iints plugin register validator validators/my_check.py

Algorithm plugins become visible in iints algorithms list. Patient model, data source, and validator plugins are registered for discovery/documentation hooks so the SDK can grow without source-code edits.

`iints plugin list`¶

Show local extension plugins.

iints plugin list
iints plugin list --kind algorithm

`iints plugin uninstall`¶

Remove a local plugin registry entry.

iints plugin uninstall "My Algorithm"
iints plugin uninstall "My Algorithm" --remove-file

`iints patientmodel list`¶

Show built-in and locally registered patient models.

iints patientmodel list

Study / Research Commands¶

`iints study-protocol`¶

Write the official study protocol bundle.

`iints run-study`¶

Run the scientific benchmark matrix.

`iints analyze`¶

Aggregate a study directory.

`iints compare-study`¶

Compare two study outputs.

`iints poster-study`¶

Generate poster-ready figures from a study.

Data Commands¶

`iints data list`¶

Show public data packs.

`iints data fetch`¶

Fetch a pack into a local directory.

`iints data research-plan`¶

Generate the curated diabetes dataset acquisition plan for local AI research.

iints data research-plan --output-dir data_packs/research_dataset_plan

Use --dataset <id> repeatedly to generate a focused plan for a subset.

`iints data certify`¶

Run data certification.

`iints data realism-check`¶

Judge whether a glucose trace looks physiologically plausible for research or demo use. Supports: - --reference free_living_t1d, --reference azt1d, or --reference hupa_ucm - --output-json results/realism_report.json - --output-html results/realism_dashboard.html - --min-realism-verdict needs_review

`iints import-carelink`, `iints import-nightscout`, `iints import-tidepool`¶

Import real-world CGM sources.

Edge / Booth Commands¶

`iints edge doctor`¶

Preflight for Raspberry Pi or UNO Q.

`iints edge quickstart`¶

Create the easiest Pi or UNO Q demo project and optionally start the Linux-side runtime.

`iints edge setup`¶

Generate an edge project scaffold.

`iints edge deploy`¶

Scaffold, upload, install, and start a Raspberry Pi edge project in one command.

`iints edge offline-bundle`¶

Build a USB-friendly offline install tarball for Raspberry Pi or UNO Q setups.

`iints edge study`¶

Run a reproducible multi-seed study directly on the current edge machine.

`iints edge long-study`¶

Run a multi-day or multi-week YAML-driven study directly on the Pi, with rolling day profiles and export-friendly nested outputs. Use --resume to continue from the next incomplete day after a reboot.

`iints edge study-snapshot`¶

Create a .tar.gz snapshot of a long-study folder for crash recovery or USB backup.

`iints edge study-export`¶

Package a long-study folder into a transfer-ready zip archive for another device.

`iints edge remote-status`, `iints edge remote-reset`, `iints edge remote-stop`¶

Run common Raspberry Pi maintenance commands remotely over SSH.

Jetson Endurance Commands¶

`iints jetson doctor`¶

Check Jetson-like hardware probes, thermal zones, and NVIDIA tooling before a long headless run.

`iints jetson endurance start`¶

Run a headless adversarial endurance study.

iints jetson endurance start \
  --algo algorithms/example_algorithm.py \
  --predictor models/lstm_predictor.pt \
  --duration 7d \
  --output-dir results/jetson_7day \
  --profile mixed_adversarial \
  --seed 42 \
  --checkpoint-interval 360 \
  --hardware-sample-interval 60

Add --wall-clock when the study horizon should consume real time instead of finishing as fast as possible. A run such as --duration 1d --wall-clock therefore lasts about 24 real hours and writes a training-ready research/ bundle next to the normal endurance outputs.

`iints jetson endurance status`¶

Show progress, current glucose, TIR so far, interventions, critical events, the latest checkpoint, resume count, and wall-clock ETA.

`iints jetson endurance monitor`¶

Print the same status repeatedly with --watch.

`iints jetson endurance stop`¶

Request a safe stop and optional report finalization.

`iints jetson endurance export`¶

Package the complete endurance folder into a transfer-ready .zip.

`iints jetson endurance finalize-research`¶

Train post-run local research models from one endurance bundle and write a held-out closed-loop evaluation report.

`iints jetson endurance install-service`¶

Write a systemd service file with automatic --resume for multi-day Jetson runs. Pass --wall-clock here too when the generated service should preserve real-time pacing.

Local AI Research Commands¶

`iints research blend-datasets`¶

Blend already prepared real datasets into one source-aware predictor dataset.

`iints research prepare-ohio`¶

Prepare a local OhioT1DM XML folder into a gitignored processed dataset. Do not commit the raw OhioT1DM-volledig/ folder to GitHub:

iints research prepare-ohio \
  --input-dir /path/to/OhioT1DM-volledig \
  --splits train \
  --output data_packs/public/ohio_t1dm_full/processed/ohio_train.csv \
  --report data_packs/public/ohio_t1dm_full/processed/ohio_train_quality_report.json

`iints research glucose-model build-dataset`¶

Normalize one or more prepared glucose datasets into the dedicated iints-glucose-forecast-v0 training contract:

iints research glucose-model build-dataset \
  --input data_packs/public/ohio_t1dm_full/processed/ohio_train.csv \
  --input results/realism_learning_10k/research/predictor_training.csv \
  --labels ohio_full,sim_10k \
  --profile long \
  --output-dir models/iints-glucose-forecast-v0/dataset

`iints research glucose-model train`¶

Train the dedicated glucose-forecast model and optionally build a Hugging Face-ready export folder:

iints research glucose-model train \
  --data models/iints-glucose-forecast-v0/dataset/glucose_training_dataset.csv \
  --config models/iints-glucose-forecast-v0/dataset/glucose_model_config.yaml \
  --output-dir models/iints-glucose-forecast-v0 \
  --epochs 220 \
  --comparison-dir results/glucose_model_comparison \
  --export-hf

`iints research glucose-model compare`¶

Compare transparent baselines and trained MSE/Band/PINN checkpoints against physiology-aware gates:

iints research glucose-model compare \
  --data data_packs/public/ohio_t1dm_full/processed/ohio_test.csv \
  --config models/iints-glucose-forecast-v0/dataset/glucose_model_config.yaml \
  --model mse=models/glucose_mse/predictor.pt \
  --model pinn=models/iints-glucose-forecast-v0/predictor.pt \
  --mc-samples 30 \
  --output-dir results/glucose_model_comparison

`iints research glucose-model export-hf`¶

Package predictor.pt, training_report.json, the model config, privacy/limitations notes, examples, comparison metrics, a research-only model card, and a redacted public dataset manifest for Hugging Face:

iints research glucose-model export-hf \
  --model-dir models/iints-glucose-forecast-v0 \
  --dataset-manifest models/iints-glucose-forecast-v0/dataset/glucose_dataset_manifest.json \
  --comparison-dir results/glucose_model_comparison \
  --repo-id IINTS/iints-glucose-forecast-v0

`iints research glucose-model jetson-train-hf`¶

Continue training an existing Hugging Face glucose model on Jetson with warm-start, candidate comparison, and local champion promotion:

iints research glucose-model jetson-train-hf \
  --repo-id IINTS/iints-glucose-forecast-v0 \
  --dataset models/iints-glucose-forecast-v0/dataset/glucose_training_dataset.csv \
  --dataset-manifest models/iints-glucose-forecast-v0/dataset/glucose_dataset_manifest.json \
  --work-dir models/jetson_hf_training \
  --max-trials 1 \
  --epochs 2 \
  --batch-size 64 \
  --upload-mode none

Use --upload-mode pr after review to upload a promoted champion as a Hugging Face pull request. The command uploads model artifacts and redacted metadata only, not raw private dataset rows.

`iints research build-control-dataset`¶

Combine one or more run bundles into a supervised controller teacher dataset.

`iints research train-controller`¶

Train the first auditable local controller baseline from safe-action labels.

`iints research train-neural-controller`¶

Train the stronger PyTorch controller from the same supervised safe-action labels.

`iints research evaluate-controller`¶

Compare a learned controller against the clinical baseline on held-out presets and seeds.

`iints research local-ai-lab`¶

Combine completed Jetson/simulator runs into one local AI workspace: predictor dataset, controller-teacher dataset, dataset card, local controller models, optional predictor training, and held-out controller evaluation.

`iints research train-local-ai`¶

Friendly alias for the same local AI workspace command. Prefer this name in new demos and docs:

iints research train-local-ai \
  --run day1=results/jetson_research_day \
  --output-dir results/local_ai_lab

Full workflow: Jetson Endurance Mode.

Results Management¶

`iints results`¶

Index all run-level results.csv files and every generated artifact under a results root:

iints results --root results

This writes a compact management bundle:

run_index.csv
artifact_inventory.csv
RESULTS_INDEX.md
result_manager_manifest.json
results_index.xlsx when spreadsheet export is available

Use --include-raw only when you want one combined long table for downstream local-AI/data analysis:

iints results --root results/research_realism_sweep_20260603_02 --include-raw

The same command is also available as iints research results-index for research workflows.

Evidence Commands¶

`iints run-doctor`¶

Preflight an algorithm, patient YAML, scenario JSON, duration, time step, and output folder before a long run:

iints run-doctor \
  --algo algorithms/example_algorithm.py \
  --patient-config-path patients/stable_patient.yaml \
  --scenario-path scenarios/clinic_safe_baseline.json \
  --duration 1440 \
  --time-step 5

It catches missing files, validation errors, aggressive glucose drift, likely pre-meal hypoglycemia, and output path problems.

`iints evidence build`¶

Build a public research evidence bundle from one or more completed runs:

iints evidence build \
  --run normal=results/live_demo/results/01_normal_run \
  --run stress=results/live_demo/results/02_meal_stress \
  --output-dir results/live_demo/evidence_bundle

Optional inputs:

--local-ai-dir results/local_ai_lab
--pump-bundle-dir bundles/pico_bench_bundle

`iints report --style agp`¶

Generate an AGP-style research PDF from a dense simulation or CGM CSV:

iints report \
  --results-csv results/one_day/results.csv \
  --style agp \
  --png \
  --svg \
  --subject-name "stable demo run" \
  --bundle-dir results/one_day/agp_report

This writes agp_report.pdf, agp_summary.json, agp_assets/agp_profile.png, agp_assets/agp_profile.svg, agp_assets/daily_profiles.png, and agp_assets/daily_profiles.svg. When a run contains explainable_events, the AGP asset folder also includes xai_events.txt for human review and xai_events.json for downstream analysis. The layout includes glucose statistics, time-in-ranges, an AGP-style modal-day percentile plot, and daily glucose profiles.

`iints safety-visualize`¶

Create a standalone HTML safety visualizer from a run CSV:

iints safety-visualize \
  --results-csv results/one_day/results.csv \
  --output-html results/one_day/safety_visualizer.html \
  --output-json results/one_day/safety_visualizer.json

Normal iints run, iints presets run, run_full(...), and run_simulation(...) already create realism_report.json, realism_dashboard.html, safety_visualizer.html, and safety_visualizer.json inside the run bundle.

Pico Pump Bench Commands¶

`iints pump init`¶

Create a bench-only Pico pump lab workspace.

`iints pump compile`¶

Package a simulated SDK algorithm with locked, non-actuating Pico firmware:

iints pump compile \
  --algorithm algorithms/pico_bench_algorithm.py \
  --output-dir bundles/pico_bench_bundle

`iints pump bench-test`¶

Validate the bundle before upload:

iints pump bench-test \
  --bundle-dir bundles/pico_bench_bundle \
  --output-json bundles/pico_bench_bundle/bench_test_report.json

`iints pump upload`¶

Copy the locked bundle to a writable Pico/CircuitPython-style drive after explicit bench-only confirmation.

FPGA Safety-Core Commands¶

`iints fpga start`¶

Run the easiest FPGA quickstart: create a lab scaffold and run the golden mock safety-core demo.

iints fpga start --output-dir results/fpga_start

`iints fpga doctor`¶

Check whether FPGA mode can run locally. Mock transport works without hardware; serial transport is optional and requires pyserial.

`iints fpga setup`¶

Create a bench-only FPGA lab workspace:

iints fpga setup --output-dir iints_fpga_lab

The workspace includes a safety contract, demo events, a golden night_hypo_risk scenario, JSON-lines protocol description, Verilog scaffold, Verilog smoke test, JSON-lines bridge stub, FPGA_STORY.md, and a mock-demo shell script.

`iints fpga simulate`¶

Run a software-reference versus FPGA-style safety-core comparison:

iints fpga simulate \
  --events iints_fpga_lab/scenarios/night_hypo_risk.json \
  --output-dir results/fpga_mock_run

Use --transport serial --port /dev/ttyUSB0 when a real FPGA bridge is available.

`iints fpga export-events`¶

Convert an existing IINTS results CSV into FPGA event JSON:

iints fpga export-events \
  --results-csv results/my_run/results.csv \
  --output-events results/fpga_events.json

`iints fpga replay`¶

Convert an existing IINTS results CSV and immediately run the FPGA comparison:

iints fpga replay \
  --results-csv results/my_run/results.csv \
  --output-dir results/fpga_replay

`iints fpga compare`¶

Read fpga_comparison.json and fail if hardware-style output diverged from the SDK reference:

iints fpga compare --run-dir results/fpga_mock_run

`iints fpga report`¶

Print the report location and summary:

iints fpga report --run-dir results/fpga_mock_run

`iints fpga demo`¶

Create the lab scaffold and run a mock FPGA safety-core demo in one command:

iints fpga demo --output-dir results/fpga_demo

The demo bundle includes reviewer-friendly top-level files: events.csv, results.json, manifest.json, and report.md.

Maker Faire Commands¶

`iints makerfaire up`¶

Start the Pi booth flow.

`iints makerfaire autostart`¶

Prepare booth autostart files.

`iints makerfaire watchdog`¶

Recover the booth runtime if it stops.

Diagnostics¶

`iints doctor`¶

Basic and full environment checks.

iints doctor
iints doctor --full --suggest
iints doctor --smoke-run

doctor reports the installed SDK version, active Python executable, install path, and available command groups. This is the fastest way to catch an old Python interpreter that silently resolves only legacy SDK releases.

`iints profiles`¶

iints profiles presets
iints profiles create --name stable_patient --preset stable-demo
iints profiles create --name endurance_patient --preset endurance

Starter presets are provided for stable-demo, stress-test, and endurance.

Full Details¶

For every option and advanced workflow, continue to: - Choose Your Path - CLI & Advanced Reference - Scientific Workflow - Study Analysis

Command Reference¶

Beginner-Friendly Entry Points¶

iints guide¶

iints start¶

iints onboard¶

iints demo¶

iints demo-live¶

iints quickstart¶

iints run --wizard¶

iints update¶

iints delete¶

Core Simulation Commands¶

iints run¶

iints run-full¶

iints run-parallel¶

iints benchmark¶

Extension Commands¶

iints plugin install¶

iints plugin register¶

iints plugin list¶

iints plugin uninstall¶

iints patientmodel list¶

Study / Research Commands¶

iints study-protocol¶

iints run-study¶

iints analyze¶

iints compare-study¶

iints poster-study¶

Data Commands¶

iints data list¶

iints data fetch¶

iints data research-plan¶

iints data certify¶

iints data realism-check¶

iints import-carelink, iints import-nightscout, iints import-tidepool¶

Edge / Booth Commands¶

iints edge doctor¶

iints edge quickstart¶

iints edge setup¶

iints edge deploy¶

iints edge offline-bundle¶

iints edge study¶

iints edge long-study¶

iints edge study-snapshot¶

iints edge study-export¶

iints edge remote-status, iints edge remote-reset, iints edge remote-stop¶

Jetson Endurance Commands¶

iints jetson doctor¶

iints jetson endurance start¶

iints jetson endurance status¶

iints jetson endurance monitor¶

iints jetson endurance stop¶

iints jetson endurance export¶

iints jetson endurance finalize-research¶

iints jetson endurance install-service¶

Local AI Research Commands¶

iints research blend-datasets¶

iints research prepare-ohio¶

iints research glucose-model build-dataset¶

iints research glucose-model train¶

iints research glucose-model compare¶

iints research glucose-model export-hf¶

iints research glucose-model jetson-train-hf¶

iints research build-control-dataset¶

iints research train-controller¶

iints research train-neural-controller¶

iints research evaluate-controller¶

iints research local-ai-lab¶

iints research train-local-ai¶

Results Management¶

iints results¶

Evidence Commands¶

iints run-doctor¶

iints evidence build¶

iints report --style agp¶

iints safety-visualize¶

Pico Pump Bench Commands¶

iints pump init¶

iints pump compile¶

iints pump bench-test¶

`iints guide`¶

`iints start`¶

`iints onboard`¶

`iints demo`¶

`iints demo-live`¶

`iints quickstart`¶

`iints run --wizard`¶

`iints update`¶

`iints delete`¶

`iints run`¶

`iints run-full`¶

`iints run-parallel`¶

`iints benchmark`¶

`iints plugin install`¶

`iints plugin register`¶

`iints plugin list`¶

`iints plugin uninstall`¶

`iints patientmodel list`¶

`iints study-protocol`¶

`iints run-study`¶

`iints analyze`¶

`iints compare-study`¶

`iints poster-study`¶

`iints data list`¶

`iints data fetch`¶

`iints data research-plan`¶

`iints data certify`¶

`iints data realism-check`¶

`iints import-carelink`, `iints import-nightscout`, `iints import-tidepool`¶

`iints edge doctor`¶

`iints edge quickstart`¶

`iints edge setup`¶

`iints edge deploy`¶

`iints edge offline-bundle`¶

`iints edge study`¶

`iints edge long-study`¶

`iints edge study-snapshot`¶

`iints edge study-export`¶

`iints edge remote-status`, `iints edge remote-reset`, `iints edge remote-stop`¶

`iints jetson doctor`¶

`iints jetson endurance start`¶

`iints jetson endurance status`¶

`iints jetson endurance monitor`¶

`iints jetson endurance stop`¶

`iints jetson endurance export`¶

`iints jetson endurance finalize-research`¶

`iints jetson endurance install-service`¶

`iints research blend-datasets`¶

`iints research prepare-ohio`¶

`iints research glucose-model build-dataset`¶

`iints research glucose-model train`¶

`iints research glucose-model compare`¶

`iints research glucose-model export-hf`¶

`iints research glucose-model jetson-train-hf`¶

`iints research build-control-dataset`¶

`iints research train-controller`¶

`iints research train-neural-controller`¶

`iints research evaluate-controller`¶

`iints research local-ai-lab`¶

`iints research train-local-ai`¶

`iints results`¶

`iints run-doctor`¶

`iints evidence build`¶

`iints report --style agp`¶

`iints safety-visualize`¶

`iints pump init`¶

`iints pump compile`¶

`iints pump bench-test`¶

`iints pump upload`¶

`iints fpga start`¶

`iints fpga doctor`¶

`iints fpga setup`¶

`iints fpga simulate`¶

`iints fpga export-events`¶

`iints fpga replay`¶

`iints fpga compare`¶

`iints fpga report`¶

`iints fpga demo`¶

`iints makerfaire up`¶

`iints makerfaire autostart`¶