Перейти к содержимому
TakoAPI
Назад к навыкам

fieldy-ai-webhook

от mrzilvis

Moltbook
1.9kЧисло загрузок по данным clawskills.sh
3Звёзды
0Просмотры
0 лайковВойдите, чтобы поставить лайк навыку
—Ещё нет оценок
Ваша оценка

Кратко

Wire a Fieldy webhook transform into Moltbot hooks.

Описание

Что делает этот навык

Configures a Moltbot Gateway webhook endpoint so incoming Fieldy app transcripts are processed by a transform script before reaching an agent. Transcripts containing the wake word "Hey, Fieldy" (or "Fieldy") trigger an agent run with the following text; all others are logged to JSONL files. Wake word behavior and logging can be customized by editing fieldy-webhook.js.

The transform layer filters transcripts by wake word before agent invocation, so the agent only runs on relevant input rather than every transcript the app sends.

Когда его использовать

  • Trigger an AI agent from a Fieldy voice transcript using a wake word
  • Log all incoming transcripts to JSONL without invoking the agent
  • Filter meeting transcripts so only voice commands reach automation
  • Customize wake word detection or parsing for a specific workflow
  • Route Fieldy webhook events to a self-hosted Moltbot gateway

Пример сценария

ВВОД

User sends: POST /hooks/fieldy with body {"transcript":"Hey Fieldy summarize this: hello world"}

АГЕНТ
  1. 1

    Gateway receives the POST request and routes it through fieldy-webhook.js

  2. 2

    Transform script checks the transcript for the wake word "Hey, Fieldy" or "Fieldy"

  3. 3

    Wake word is detected, so the script extracts the text after it: "summarize this: hello world"

  4. 4

    Transform passes the extracted message to the agent via the configured messageTemplate

  5. 5

    Agent runs with the input "summarize this: hello world"

ВЫВОД

Agent executes with the post-wake-word text as its prompt

Требования

  • Fieldy app account with access to Developer Settings
  • Moltbot hooks token to set in moltbot.json

Установка

clawhub install mrzilvis/fieldy-ai-webhook

Выполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.

Ссылки