AGILab Logo
Docs version 92a59c9

Start

  • Newcomer Guide
    • Shortest first proof
    • What success looks like
    • What to ignore on day 1
    • The first 10 minutes
    • The core ideas, in plain language
    • Common newcomer traps
    • Only after the first proof
    • Where to go next
  • Quick-Start
    • Prerequisites
      • Recommended first proof path
      • If the first proof fails
      • Alternative install routes
      • Private apps or framework contributor setup
      • Run without PyCharm (CLI wrappers)
      • Optional developer workflow
      • Codex workflow
      • Cluster installs
      • Next steps
      • Support
      • License
  • Newcomer First-Failure Recovery
    • Before anything else
    • Failure 1: uv is missing
    • Failure 2: ./install.sh --install-apps --test-apps fails
    • Failure 3: the built-in app path is not found
    • Failure 4: the web UI or About/ORCHESTRATE smoke fails
    • Failure 5: no fresh output appears under ~/log/execute/flight/
    • When you are past the newcomer hurdle
  • Compatibility Matrix
    • Current public matrix
    • How to use this matrix
    • What remains roadmap work

Use

  • Introduction
    • What AGILab is
    • Historical note
    • Why the project is built this way
    • Main dependencies
    • What to read next
  • Features
    • agi-core
    • agilab
  • Notebook Quick Start
    • When this path helps
    • What this path does not cover
    • Repository launch flow
    • Minimal notebook cells
    • How this maps back to the GUI
    • Optional next step
    • Related pages
  • AGILab
    • Reference
      • main()

Build

  • AGILab Architecture
    • Component view
    • Pipeline example
    • agilab.py navigation
    • Manager vs worker responsibilities
    • Layers at a glance
    • Runtime flow
    • Repository map
    • Core vs optional apps
    • Documentation map
    • See also
  • AGI Core Architecture
    • Modules at a glance
    • Execution flow
    • Repository pointers
    • Tips for contributions
    • See also
  • AGILab in the MLOps Toolchain
    • Where AGILab helps
    • What AGILab does not aim to cover
    • Positioning vs. other tools
    • Framework comparison
    • Selection guide
    • Suggested workflow
    • See also
  • Learning Workflows
    • Training vs inference
      • Reinforcement learning (example)
      • Optimization / ILP baseline (example)
    • Continuous learning (optional)
    • Graph neural networks (message passing)
    • Federated learning (optional)
  • Framework API
    • Core packages
      • agi-env API
        • Usage Example
        • Reference
      • agi-node API
        • Path handling
        • Argument helpers
        • Managed PC path remapping
        • Output directory helpers
        • Reference
      • agi-distributor API
        • Understanding modes_enabled and mode
        • Usage Example
        • Reference
    • Working with the API
    • App structure conventions
  • Pinned public framework submodule
    • Recommended default
    • Why this is the default
    • Supported resolution order
    • Local workflow
    • Consequences
  • Cluster
    • Overview
    • Principle
    • Normal AGILAB Workflow
    • SSH key setup
      • Distributed Workers
        • Prerequisites
        • Step 1: Configure Distributed Execution in ORCHESTRATE
        • Step 2: Let ORCHESTRATE Generate the Snippet
        • Reading mode and modes_enabled
        • Quick UI Walkthrough
        • Equivalent Generated Snippets
        • Step 3: Validate the Distribution Before Running
        • Step 4: Reuse the Generated Snippet in PIPELINE
        • Best Practices
        • Troubleshooting
      • Key Generation
        • 1. Generate the keys
        • 2. Loading the private key in SSH Agent
        • 3. Copy the public key to the server
        • Troubleshooting
        • Useful links
  • Distributed Workers
    • Prerequisites
    • Step 1: Configure Distributed Execution in ORCHESTRATE
    • Step 2: Let ORCHESTRATE Generate the Snippet
    • Reading mode and modes_enabled
    • Quick UI Walkthrough
    • Equivalent Generated Snippets
    • Step 3: Validate the Distribution Before Running
    • Step 4: Reuse the Generated Snippet in PIPELINE
    • Best Practices
    • Troubleshooting
  • Module Reference
    • Canonical pages
  • Environment Variables
    • Security note
  • FAQ
    • Missing worker packages during AGI.run_*
    • Why installers still build eggs
    • Do we already have DAG/task orchestration?
    • Who manages multithreading when Dask is disabled?
    • Regenerating IDE run configurations
    • Using run configurations without PyCharm
    • “VIRTUAL_ENV … does not match the project environment” warning
    • Why does a run create distribution.json?
    • Switching the active app in the web interface
    • Docs drift after touching core APIs
    • AGI.install_* fails looking for pyproject.toml
    • Where are installer logs written?
    • Why did my local coverage run not change the README badge?
    • Why can agi-gui be at 99% while global agilab coverage is lower?
    • Which docs repo should I edit?
    • What does tools/newcomer_first_proof.py actually prove?
  • Project Files Structure
  • Troubleshooting
    • A - Prerequisite:
    • B - Run/Debug configurations (PyCharm and shell):
    • C - Exemple of Tests Sequence:
    • D - agilab_run_dev vs agilab_run_enduser vs lab_run:
    • E - macOS NFS server checklist:
  • Known Bugs
    • <install.sh> do not install your Run/Debug Configurations :
    • <install.sh> freeze:
    • <UV> VIRTUAL_ENV Warning
    • <Python> Python
    • <UV> Sync Failed
    • <DASK> Debug Issue
    • <PYCHARM> Run/Debug Configuration is Broken
    • <PYCHARM> Can’t open your project
    • Failed to read pydantic metadata
  • Licenses
    • Agi-cluster Licenses
    • Agi-node Licenses
    • Agi-env Licenses
    • Agi-core Licenses
    • Mycode-project Licenses
    • Flight-project Licenses

Pages

  • About AGILab
    • How pages are presented
    • Core pages
    • Core page tour
    • Page bundles (apps-pages)
    • First-time navigation
    • See also
  • PROJECT
    • Page snapshot
    • Sidebar
    • Main Content Area
    • Troubleshooting and checks
    • See also
    • Support
  • ORCHESTRATE
    • Introduction
    • Page snapshot
    • Sidebar
    • Main Content Area
    • Execution Mode Values
    • From UI to Snippet Fields
    • Distributed Workflow
    • Snippet Handoff to Pipeline
    • Service Mode Health
    • Service health JSON export
    • Troubleshooting and checks
    • See also
  • PIPELINE
    • Page snapshot
    • Sidebar
    • Main Content Area
      • ASSISTANT
      • MLflow tracking
      • HISTORY
    • Troubleshooting and checks
    • See also
  • ANALYSIS
    • Introduction
    • Page snapshot
    • Sidebar
    • Main Content Area
    • Tips & Notes
    • Troubleshooting and checks
    • See also

Service and Operations

  • Service Mode
    • When to use it
    • Fast path in ORCHESTRATE (web interface)
    • Action semantics
    • End-to-end CLI example
    • SLA thresholds
    • Operational checks
    • Common pitfalls
    • Related pages
  • Service install paths
    • Terminology
    • Key Files And Environment Variables
    • How App Symlinks Are Resolved
    • Cleaning Up Broken Links
    • Practical Checklist
  • Service Health JSON Schema
    • Overview
    • Core fields
    • Optional fields
    • Minimal monitoring rule
    • CLI example

Examples

  • Demos
    • AGILAB Video Tutorial And Slideshow Guide
      • Recommended tutorial package
      • Which format to use
      • Fastest live workflow
        • Three-project technical hero demo
      • Self-generated fallback
      • Storyboard
        • Flight 30-second version
        • Flight 45-second version
        • Meteo forecast 45-second version
        • Flight 60-second version
        • Flight 3-minute version
        • UAV queue 45-second version
        • UAV Relay Queue 3-minute version
      • Slideshow structure
      • Recording and visual rules
      • Quality checklist
      • Default tagline
    • Start Here
    • Which Demo To Watch
    • Core Tour
    • What The Reels Show
    • See Also
  • Execution Playground
    • What is included
    • Where you see it in the UI
    • Why this example matters
    • What the benchmark shows
    • Measured local benchmark
    • 2-node 16-mode matrix
      • Mode families
      • How to read the matrix quickly
      • execution_pandas_project
      • execution_polars_project
    • How to run it
    • What to look for
  • Notebook Migration Example
    • Repository material
    • Why migrate
    • Migrated pipeline shape
    • Real built-in project
    • ANALYSIS page
    • Suggested migration path
  • MyCode Project
    • Overview
    • Scientific placeholders
    • Manager (mycode.mycode)
    • Args (mycode.app_args)
    • Worker (mycode_worker.mycode_worker)
    • Assets & Tests
    • API Reference
      • Mycode
        • Mycode.__init__()
        • Mycode.as_dict()
        • Mycode.build_distribution()
        • Mycode.from_toml()
        • Mycode.pool_init()
        • Mycode.stop()
        • Mycode.to_toml()
        • Mycode.work_done()
        • Mycode.work_pool()
        • Mycode.worker_vars
      • MycodeApp
      • ArgsModel
      • ArgsOverrides
      • MycodeArgs
        • MycodeArgs.data_in
        • MycodeArgs.data_out
        • MycodeArgs.files
        • MycodeArgs.model_config
        • MycodeArgs.nfile
        • MycodeArgs.nread
        • MycodeArgs.nskip
        • MycodeArgs.reset_target
      • MycodeArgsTD
        • MycodeArgsTD.data_in
        • MycodeArgsTD.data_out
        • MycodeArgsTD.files
        • MycodeArgsTD.nfile
        • MycodeArgsTD.nread
        • MycodeArgsTD.nskip
        • MycodeArgsTD.reset_target
      • dump_args()
      • ensure_defaults()
      • load_args()
      • merge_args()
      • MycodeWorker
  • Flight Project
    • Overview
    • Scientific highlights
    • Manager (flight.flight)
    • Args (flight.flight_args)
    • Worker (flight_worker.flight_worker)
    • Assets & Tests
    • API Reference
      • Flight
        • Flight.__init__()
        • Flight.as_dict()
        • Flight.build_distribution()
        • Flight.extract_plane_from_file_name()
        • Flight.from_toml()
        • Flight.get_data_from_files()
        • Flight.get_data_from_hawk()
        • Flight.get_partition_by_planes()
        • Flight.ivq_logs
        • Flight.to_toml()
      • FlightApp
      • ArgsModel
      • ArgsOverrides
      • FlightArgs
        • FlightArgs.data_in
        • FlightArgs.data_out
        • FlightArgs.data_source
        • FlightArgs.datemax
        • FlightArgs.datemin
        • FlightArgs.files
        • FlightArgs.model_config
        • FlightArgs.nfile
        • FlightArgs.nread
        • FlightArgs.nskip
        • FlightArgs.output_format
        • FlightArgs.reset_target
        • FlightArgs.sampling_rate
        • FlightArgs.to_toml_payload()
      • FlightArgsTD
        • FlightArgsTD.data_in
        • FlightArgsTD.data_out
        • FlightArgsTD.data_source
        • FlightArgsTD.datemax
        • FlightArgsTD.datemin
        • FlightArgsTD.files
        • FlightArgsTD.nfile
        • FlightArgsTD.nread
        • FlightArgsTD.nskip
        • FlightArgsTD.output_format
        • FlightArgsTD.reset_target
        • FlightArgsTD.sampling_rate
      • apply_source_defaults()
      • dump_args()
      • dump_args_to_toml()
      • ensure_defaults()
      • load_args()
      • load_args_from_toml()
      • merge_args()
      • FlightWorker
        • FlightWorker.calculate_speed()
        • FlightWorker.pool_init()
        • FlightWorker.pool_vars
        • FlightWorker.preprocess_df()
        • FlightWorker.start()
        • FlightWorker.stop()
        • FlightWorker.work_done()
        • FlightWorker.work_init()
        • FlightWorker.work_pool()

Reference

  • AGILab project
  • AGILab components

Roadmap

  • Roadmap
    • AGILab future work
      • Recommended near-term execution order
      • Streamlit-inspired AGILab views
        • 0. First-proof wizard
        • 0b. Run manifest + evidence bundle
        • 1. Experiment Cockpit
        • 2. Evidence / Release View
        • 3. Scenario Playback View
        • 4. Realtime Analytical and Geospatial Views
        • 5. Run Diff / Counterfactual Analysis
      • Backend observability and audit architecture
        • 1. Elastic or OpenSearch + Grafana
        • 2. OpenSearch + OpenSearch Dashboards
        • 3. Postgres + Superset
      • Connectors and integration
        • 1. Connector framework hardening
      • Distributed execution and reduction
        • 1. First-class reduce contract
        • 2. External system connectors
        • 3. Connector-aware views
        • 4. DeepWiki/Open-style repository knowledge layer
      • Decision guidance
      • Final consolidated poll
        • Comment template for issues/2
        • Current candidate priorities
      • Reference URLs
    • Feature: versioned pipeline step templates
      • Problem
      • Current product stance
      • Proposal
      • Example shape
      • Expected behaviour
      • Why this is better
      • Transition plan
        • Phase 1
        • Phase 2
        • Phase 3
        • Phase 4
      • Non-goals
      • Product impact
AGILab
  • AGILab project
  • View page source

AGILab project

  • AGILab on GitHub

AGILab components

  • agi-env on PyPI

  • agi-cluster on PyPI

  • agi-node on PyPI

  • agilab on PyPI

Previous Next

© Copyright 2026, THALES SIX GTS France SAS.

Built with Sphinx using a theme provided by Read the Docs.