メニュー
Bootstrap Cognition — automatically installs Screenpipe, sets up auth, connects everything, and begins your first learning session. One command, zero config.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
Connect a notification channel (Slack, Gmail, etc.) so Cognition can send you learning exercises throughout the day — even when your laptop is off.
Schedule learning exercises for the next 24 hours via Slack/Gmail. Runs at the end of every session so exercises arrive even when Claude Code is closed. The API decides WHEN and WHAT technique — you build the exercise and schedule delivery.
Send screen observations to the Cognition API as learning events. Transforms Screenpipe data into API event format.
Launch the Cognition GUI app. Opens the Electron window and connects Claude Code to it via WebSocket. Use this after /cognition:start or whenever you need the GUI.
Generate and deliver a learning exercise based on what the user is forgetting. Orchestrates different learning techniques.
Test if the user's confidence matches their actual knowledge. Use when calibration ECE is high or strategy is "overconfident".
SOC 職業分類に基づく
| name | start |
| description | Bootstrap Cognition — automatically installs Screenpipe, sets up auth, connects everything, and begins your first learning session. One command, zero config. |
| disable-model-invocation | false |
Set up everything automatically. The user should not have to do anything manually.
First check if Screenpipe is already running:
curl -s http://localhost:3030/health 2>/dev/null || echo "NOT_RUNNING"
If NOT running, start it automatically in the background:
npx screenpipe@latest record > /dev/null 2>&1 &
Wait 5 seconds for it to initialize, then verify:
sleep 5 && curl -s http://localhost:3030/health 2>/dev/null || echo "STILL_NOT_RUNNING"
If it's still not running after 10 seconds, try once more. If it fails again, tell the user:
Screenpipe couldn't start automatically. Run
npx screenpipe@latest recordin a separate terminal, then come back.
If running, move on.
Check if the screenpipe MCP tools are available by trying to use them. If they're not available:
claude mcp add screenpipe -- npx -y screenpipe-mcp
Note: if MCP was just added, tell the user to restart this Claude Code session once for it to take effect. This only happens on first setup.
Check if ~/.cognition/token exists:
cat ~/.cognition/token 2>/dev/null || echo "NO_TOKEN"
If no token exists, create one automatically for the user:
mkdir -p ~/.cognition
echo "dev-secret" > ~/.cognition/token
(In production, this would open a browser for OAuth signup. For now, use the dev key.)
TOKEN=$(cat ~/.cognition/token)
curl -s -H "x-api-key: $TOKEN" https://cognition-api.fly.dev/healthz
If you get {"status":"ok"}, you're connected. If not, tell the user the API might be down and to try again in a minute.
Use the screenpipe MCP tools to search recent screen content (last 30 minutes):
Use screenpipe search-content with content_type "ocr" and limit 10
Summarize what the user has been doing. Then say:
I can see you've been working on [summary]. I'm now tracking what you're learning.
Here's what I'll do:
- Watch what you read and code via Screenpipe
- Extract key concepts and model your knowledge
- Quiz you on things you're about to forget
Commands:
/cognition:status— see your knowledge dashboard/cognition:session— run a study session/cognition:predict— check what's decaying/cognition:learn— practice right now
Then automatically ingest the recent screen content as screen_capture events via the API (use the ingest skill flow). Confirm:
Ingested [N] screen captures. Extracted [M] concepts. Your learning model is live.
After running /cognition:start, the user should have:
All of this should happen automatically with ZERO user action beyond typing /cognition:start.