Quickstart

After installation, open your application repository. Replace /path/to/adlc5 with the distribution clone or installed plugin root.

Initialize the workspace

/path/to/adlc5/scripts/init-workspace.sh --project .
/path/to/adlc5/scripts/init-feature.sh --feature my-feature --interaction hitl

Initialization creates repository context and a local .adlc5/my-feature/ tree. Review the draft .agents/ constitution so architecture, boundaries, and commands describe your application. Generated .agent-cache/ intelligence and .adlc5/ lifecycle state stay untracked.

Read Intelligence to review the generated guidance, check map freshness, and understand how story context reaches each agent.

Start delivery

In your agent host, invoke:

@adlc5 for my-feature

Describe the intended behavior, constraints, and acceptance evidence. ADLC5 captures requirements, chooses proportionate depth, plans the change, creates bounded tasks, then builds and verifies it. Resolve unclear requirements before implementation. See the lifecycle for outputs and gates.

For autonomous progression, initialize with --interaction autonomous instead of --interaction hitl. Autonomous work still obeys required checks and approval gates; it requires an agent host to execute judgment and coding actions.

Inspect progress

Run these from the application repository:

/path/to/adlc5/scripts/adlc5 pilot --feature my-feature --workspace .
/path/to/adlc5/scripts/adlc5 gate --feature my-feature --gate pr-ready --workspace .

pilot reports the next enabled action. A readiness gate can fail while work is in progress; read its missing evidence instead of editing state to claim completion.

Try the framework's consumer example

From the distribution repository, run the baseline consumer regression:

python3 dogfood/consumer/regression.py

The consumer guide explains isolated evaluation tasks. Fixture checks do not establish comparative coding quality or live host qualification.