PyAutoDNA

Specify. Compare. Evolve.

DNA records software stacks and the environments actually running them. It keeps upstream availability, package declarations, tested configurations, recommendations and observations separate. Brain coordinates changes; Heart supplies validation; Hands owns releases; Nerves owns runtime compatibility. DNA never installs packages or submits compute during collection, rendering or campaign actions.

Requires Python 3.12+; python -m pip install -e .. From this repository:

python -m dna collect local --repo ../PyAutoBrain
python -m dna board --brain ../PyAutoBrain --output _site
python -m dna show

Collection prints a private receipt and stores it in ignored private/records/. Do not upload its console output: it includes paths and distribution origins. --probe-backend explicitly initializes JAX to identify the actual backend; without it backend remains unknown. Collect on RAL or CI using their actual interpreter, then transfer the receipt privately and python -m dna import receipt.json. No remote observation is inferred from a declaration or local interpreter.

See operations, contracts and the JAX incident. Checked-in environments are unobserved until a receipt exists. Pages refresh alone cannot observe remote stacks.

The dashboard follows Brain's shared presentation standards and publishes the standard state.json, board.json and badge.json feeds. Its source includes reviewed observations of local development and the RAL shared environment; CPU, GPU, CI and release validation remain unknown until their own receipts arrive. support.yaml records upstream Python/JAX support horizons with source links and a policy review date. These dates do not certify PyAuto compatibility.