Channels
Stop checking things — events come to you.
TUTOR WITH THEFOCUS.AI
Copy this prompt into Claude, ChatGPT, or any external AI assistant. It points the assistant to the course instructions and links it to your student profile to track your progress and customize observations.
You are not enrolled yet. Enroll to generate a Student ID to track lesson completions and store learning notes.
Events push in, Claude responds
Up to this point, everything has been pull-based. You ask Claude a question, Claude responds. Channels invert this. Events push into your Claude session, and Claude responds without you asking.
A Channel connects an external source — Telegram, Discord, iMessage, or your own webhook endpoint — to a running Claude Code session. Under the hood a channel is an MCP server that pushes events instead of waiting to be called. Anthropic ships channel plugins for the common cases; start one with the --channels flag:
claude --channels plugin:telegram@claude-plugins-official
Claude sees each incoming event, understands it, and acts using every tool and MCP server already connected.
Before Channels: You check QuickBooks, you check email for invoices, you check the supplier portal. You pull information to Claude.
After Channels: A new invoice arrives → webhook fires → Claude receives it → Claude categorizes it using your skill → Claude enters it via MCP. You get a summary on Telegram.
| Channel Source | What Pushes In | What Claude Can Do |
|---|---|---|
| Telegram / Discord | Messages from you or your team | Respond conversationally, execute commands, report status |
| iMessage | Texts from your phone | Same session, no extra app — delegate by text message |
| Webhook endpoint | Any HTTP POST payload | Parse the payload, take action — Stripe events, form submissions, invoice arrivals |
| Monitoring (via webhook) | Sentry error alerts, CI failures | Analyze the error, check recent deploys, find the commit, draft a fix |
“The shift is from pulling information to having it pushed to you. Once you set up Channels, you stop checking things. Things come to you.” — Damien Galarza
Cornwall Market: a webhook receives notification when a new invoice PDF arrives in the shared email. Claude is alerted, downloads the attachment, reads the invoice, applies the categorization skill, and enters the data via QuickBooks MCP. Sarah gets a Telegram message: “New invoice from Chen’s Produce, $1,240, categorized to 5100. Approved?”
Check your understanding
Answer in your own words — write it down before opening the key. Your tutor grades against the criteria and generates fresh variants on retries.
Q1. Channels invert the model. State the inversion, then rewrite one recurring check from your own week in before/after form.
Answer key — attempt every question first
Answer key
Q1
Model answer: Inversion: from you pulling information (checking dashboards, opening inboxes) to events pushing into Claude, which acts and reports. Before: ‘every morning I check X.’ After: ‘when X changes, a webhook fires, Claude applies the rules and messages me the outcome plus anything needing approval.’
Pass criteria: pull-to-push stated; the before/after uses a real recurring check with Claude acting, not just notifying
Next: Routines