Dr-Claude-Code: Command for Claude Code

Dashboard Sync

Sync experiment metadata, activity logs, and artifact references to the live Research Dashboard.

When to run

• After uploading any artifact to HuggingFace (so it appears in the Artifacts tab)
• After writing or updating experiment notes or activity logs
• After the harvest phase of an experiment pipeline
• After any change to experiment state (phase transitions, go/no-go decisions)
• When the user says "sync the dashboard", "update the website", or "refresh the dashboard"

Step 1: Check HUGGINGFACE_REPOS.md (if syncing a specific experiment)

If an experiment name was passed as an argument, read its artifact registry:

cat notes/experiments/<experiment_name>/HUGGINGFACE_REPOS.md 2>/dev/null

This confirms what HF datasets should appear in the Artifacts tab after sync.

Step 2: Run import_experiments.py

This reads all experiment data from notes/experiments/ — including configs, READMEs, activity logs, and HF repo lists — and uploads the aggregated data to the RESEARCH_DASHBOARD HF dataset.

cd tools/visualizer && .tools-venv/bin/python scripts/import_experiments.py

If the .tools-venv doesn't exist or the command fails:

# Fallback: use system python if venv is missing
cd tools/visualizer && python3 scripts/import_experiments.py

Capture the output. If it reports errors (missing files, HF auth failure), report them to the user.

Step 3: Trigger dashboard refresh

Tell the live HF Space to re-download the updated data:

# Dashboard URL is in .raca/config.yaml
DASHBOARD_URL=$(python3 -c "import yaml; print(yaml.safe_load(open('.raca/config.yaml'))['dashboard']['url'])" 2>/dev/null)

if [ -n "$DASHBOARD_URL" ]; then
    curl -s -X POST "${DASHBOARD_URL}/api/experiments/sync"
else
    echo "No dashboard URL configured — skipping remote refresh"
fi

Step 4: Report what was synced

Tell the user:

Dashboard synced:
  Experiments: N experiments updated
  Artifacts: list any new HF datasets found (from HUGGINGFACE_REPOS.md)
  URL: <dashboard_url> (if configured)