Beta Readiness

AGILAB’s public beta claim is intentionally narrow. The release is ready for a beta promotion only when the local release gates, public documentation, and GitHub evidence agree.

Scope

The beta scope covers:

  • local reproducible execution

  • Dask-based distributed execution

  • Streamlit UI workflows for trusted operators

  • package-mode install and first-proof evidence

  • notebook and Quarto report handoff

  • MLflow tracking handoff

  • public Hugging Face demo evidence

The production claim remains experimental. Shared workspaces, exposed UIs, remote workers, credentials, sensitive data, and regulated workflows still need a deployment threat model, auth/TLS controls, secrets management, and organization-specific validation.

Required local gates

Before release, run the release-preflight parity set:

uv --preview-features extra-build-dependencies run python tools/workflow_parity.py \
  --profile agi-env \
  --profile agi-core-combined \
  --profile agi-gui \
  --profile docs \
  --profile installer \
  --profile shared-core-typing \
  --profile ty-typing \
  --profile dependency-policy

Then run the first-proof and publish rehearsal checks:

uv --preview-features extra-build-dependencies run python tools/newcomer_first_proof.py --with-run
uv --preview-features extra-build-dependencies run python tools/pypi_publish.py --repo testpypi --dry-run --verbose

For a final promotion, include the public demo smoke:

uv --preview-features extra-build-dependencies run python tools/hf_space_smoke.py --json

Readiness command

Use the gate summary before deciding:

uv --preview-features extra-build-dependencies run python tools/beta_readiness.py --json

For the final public gate:

uv --preview-features extra-build-dependencies run python tools/beta_readiness.py --final --include-network --json

Release boundary

The release should not be promoted when:

  • the working tree is dirty

  • HEAD does not match origin/main

  • release-preflight profiles drift from the PyPI publisher workflow

  • public docs disagree with the maturity snapshot

  • package classifiers drift from the intended public status

  • GitHub guardrail or coverage workflows fail on the release commit

Passing the beta gate means AGILAB is suitable for the documented trusted-operator experimentation and validation workflows. It does not mean AGILAB is a complete production MLOps platform.