The plugin works out of the box in local mode — scoring runs locally with zero config.
This command upgrades to cloud mode for RLM adaptive weight learning, team dashboards, and cross-session analytics.
What to do
-
Check current mode:
- If
CONTEXTSTELLAR_API_KEYis set → already in cloud mode, test connectivity - If not set → explain that local mode is active and working, then offer cloud upgrade
- If
-
Explain the two modes:
- Local mode (current): Scoring runs on your machine via bundled scorer.js. Zero network calls. No signup needed.
- Cloud mode (optional): Scores via API with learned weights. Enables RLM flywheel,
/contextstellar:stats,/contextstellar:optimize, and the web dashboard.
-
If the user wants cloud mode, guide them:
- Go to https://contextstellar.com/connect
- Create a project (if they don't have one)
- Generate an API key (starts with
cs_live_) - Copy the Project ID
-
Set environment variables. Tell the user to add these to their shell profile or
.env:
export CONTEXTSTELLAR_API_KEY="cs_live_your_key_here"
export CONTEXTSTELLAR_PROJECT_ID="your_project_id_here"
# Optional: custom instance URL (defaults to https://contextstellar.com)
# export CONTEXTSTELLAR_BASE_URL="https://your-instance.contextstellar.com"
- Verify connectivity:
curl -s "${CONTEXTSTELLAR_BASE_URL:-https://contextstellar.com}/api/v1/hooks/shell-stats" \
-H "Authorization: ${CONTEXTSTELLAR_API_KEY}"
If you get a JSON response with score, grade, etc., cloud mode is active.
- Confirm: After setup, the user gets:
- Automatic RLM weight learning (improves after ~100 events)
/contextstellar:stats— scoring dashboard in terminal/contextstellar:optimize— pattern-based recommendations- Web dashboard at contextstellar.com/dashboard/mission-control