HTTP and API fetch in automations

HTTP fetch lets an automation pull data from a URL you control — JSON APIs, simple text endpoints, internal tools — and feed the response into an agent run. It is how you bridge “our system has an HTTP API” to Telegram without building a custom integration for every vendor.

What it is

A Pro-oriented building block: during a scheduled or triggered automation, EasyClaw can request a URL, respect size and timeout guardrails, and pass the result into the model context so the output you see in Telegram is grounded in live data.

How it works

1. You configure fetch settings in the dashboard (URL, method where supported, headers if the product exposes them).
2. When the automation fires, the runtime performs the request inside strict budgets (duration, size).
3. The agent summarizes or acts on the payload, then delivers to your linked Telegram chat if that is how the automation is set up.

This is not a public HTTP API for EasyClaw itself — it is an integration primitive inside your automations. For pushing into EasyClaw from outside, see Inbound webhooks.

When to use it

• You have a stable internal JSON endpoint (status, metrics, queue depth) and want a daily digest in Telegram.
• You want to combine fetch output with Daily Control or Monitoring-style logic.
• You prefer one HTTPS call over maintaining a long-lived SDK integration.

Examples (patterns, not live URLs)

• “Every morning, GET our /api/deploy-status and tell me if anything is red.”
• “On each run, fetch the JSON feed our team already exposes and summarize new entries.”

Limits and plans

HTTP fetch for automations is a Pro capability in the V3 product direction. Starter remains strong for Calendar, RSS, and Telegram-native workflows. Confirm your entitlements in Starter vs Pro and checkout.

Requests are subject to timeouts and body size limits so shared infrastructure stays safe; exact ceilings are enforced in the product rather than documented as fixed numbers here.

Common mistakes

• Pointing at endpoints that require interactive login — use service tokens or machine-to-machine auth your team controls.
• Expecting fetch to bypass plan gates; if you are on Starter, use RSS or Calendar-first patterns.
• Storing secrets in chat — configure credentials in the dashboard where the product supports it.

Related

• Inbound webhooks
• Integrations overview — and Google Calendar when your automation needs agenda context.
• Dashboard overview

Next step: from the dashboard, open Automations and add or edit an automation that matches your plan; use Troubleshooting if a fetch fails repeatedly.