Agent Loop Dashboard

status idle · no state file · waiting for the first run of persistent-agent-loop.py

Plan

#StepStateTriesResult
(empty — the loop has not run yet)

Memory

KeyValue
(empty)

Journal

—idleno checkpoints yet. Checkpoints are written before and after every step when the loop runs.

How this view comes alive

Run the loop from the repo root, then regenerate this dashboard into the served tree:

python3 agents/persistent-agent-loop.py run --goal "your goal" \
    --plan "step A;;step B;;step C" \
    --cmd "python3 path/to/decision-model.py"
python3 agents/persistent-agent-loop.py dashboard \
    --out pages/agent-dashboard.html

The loop's brain is .agent/state.json (local-only, gitignored). The dashboard command also writes .agent/dashboard.html by default; the --out flag rewrites this served page so the Twin's cognitive-twin panel shows live state.