NorthwoodsRhetoric

Customization

Before executing, check for user customizations at: ~/.claude/PAI/USER/SKILLCUSTOMIZATIONS/NorthwoodsRhetoric/

If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.

Voice Notification

curl -s -X POST http://localhost:31337/notify \
  -H "Content-Type: application/json" \
  -d '{"message": "Transforming response through rhetoric module"}' \
  > /dev/null 2>&1 &

NorthwoodsRhetoric — Output Transform

Calls the deployed Northwoods Pack Worker (one Worker, five modules) to transform a draft AI response before showing it to the user.

Workflow

1. Read pack.url and auth.token from ~/.claude/PAI/USER/SKILLCUSTOMIZATIONS/NorthwoodsPack/CONFIG.md.
2. POST the draft response text + context hint to {pack.url}/rhetoric/transform.
3. If the Worker returns transformed text, use it as the final response.
4. If the Worker is unreachable (network down, 5xx error), return the original draft and log a quiet warning.

Request

PACK_URL="<from CONFIG.md pack.url>"
TOKEN="<from CONFIG.md auth.token>"

curl -s -X POST "$PACK_URL/rhetoric/transform" \
  -H "Content-Type: application/json" \
  -H "X-Pack-Token: $TOKEN" \
  -d '{
    "text": "the draft response markdown",
    "context": "chat"
  }'

Response shape

{
  "text": "transformed response",
  "applied": ["strip-emoji-headers", "split-long-sentences", ...],
  "warnings": []
}

Use the text field as the final response. The applied and warnings arrays are diagnostic — surface only if a warning fires.

Context values

Value	When
chat	Conversational reply, Discord, Slack, casual back-and-forth
doc	Long-form writing, blog posts, reports
code-explain	Explaining code — preserves code blocks
reflection	Inner-voice/mirror work — preserves formatting hierarchy