Claude Code Channels Launches: Control Your AI Agent from Telegram and Discord

Key Takeaways

Anthropic announced Claude Code Channels on March 20, 2026 as a research preview feature, introducing a plugin-based system that lets developers send messages from Telegram or Discord directly into a running Claude Code session on their local machine. The session processes requests using full filesystem, MCP, and git access, then replies through the same messaging app. This creates a two-way bridge between messaging platforms and a developer's working environment, representing a fundamental shift from synchronous terminal interaction to asynchronous, mobile-friendly AI agent control.

The feature requires Claude Code v2.1.80 or later and is available to Pro and Max individual users by default. Team and Enterprise organizations must have an admin explicitly enable channels in managed settings before members can use them.

Feature Overview

1. Two-Way Messaging Bridge

Channels work through MCP (Model Context Protocol) servers that push events into your running Claude Code session. Unlike integrations that spawn fresh cloud sessions or poll for updates, channel messages arrive in the session you already have open with your real files and project context. When Claude replies through a channel, the actual response appears on Telegram or Discord, while your terminal shows the tool call and confirmation.

This means a developer can be away from their desk, send a message from their phone asking Claude to check the status of a failing test, and Claude will execute the work locally against the actual codebase and reply in the same chat thread.

2. Plugin Architecture

Channels use a plugin-based design that makes the system extensible. Telegram and Discord are the two officially supported platforms in the research preview, but the architecture is designed so additional platforms can follow. Each channel is installed as a plugin via the /plugin install command and configured with platform-specific credentials (a BotFather token for Telegram or a Discord Developer Portal bot token).

The community has already begun requesting Slack, WhatsApp, and iMessage support, and Anthropic has published the full plugin source code on GitHub under the claude-plugins-official repository for developers who want to build custom channels.

3. Security: Sender Allowlists and Pairing

Every approved channel plugin maintains a sender allowlist. Only IDs that have been explicitly paired and approved can push messages into a session. All other senders are silently dropped. The pairing process requires physical access to both the messaging platform and the Claude Code terminal: a user sends a message to the bot, receives a pairing code, and enters that code in their Claude Code session.

The allowlist also controls permission relay. If Claude encounters a permission prompt while the developer is away, channels with the relay capability can forward these prompts to the messaging app for remote approval or denial. Only allowlisted senders should be trusted with this authority.

4. Enterprise Controls

For Team and Enterprise plans, channels are disabled by default and require explicit admin enablement through claude.ai Admin settings or managed settings. This gives organizations control over whether their developers can use the feature before it reaches general availability. Individual Pro and Max users can opt in per session using the --channels flag.

5. Competitive Context: The OpenClaw Response

The timing of Claude Code Channels is significant. OpenClaw, an open-source agentic AI platform, gained massive traction in March 2026 by enabling developers to manage fleets of AI agents through messaging apps like WhatsApp and Telegram from their home computers. VentureBeat described Claude Code Channels as an "OpenClaw killer" that brings similar mobile agent control capabilities into Anthropic's ecosystem. While OpenClaw offers a fully open-source, multi-model approach, Claude Code Channels integrates directly with Anthropic's models and the full Claude Code development environment.

Usability Analysis

The setup process for Claude Code Channels requires moderate technical knowledge. Users must create a bot on their chosen platform (BotFather for Telegram, Discord Developer Portal for Discord), install the plugin, configure credentials, restart Claude Code with the --channels flag, and complete the pairing process. This is straightforward for experienced developers but represents a barrier for casual users.

The practical value proposition is strongest for developers who spend significant time away from their primary workstation but want their Claude Code agent to remain productive. Common scenarios include reviewing CI results, triggering code fixes, or asking questions about a codebase from a mobile device. The webhook receiver capability also enables event-driven workflows where CI failures, error tracker alerts, or deploy pipeline events arrive directly in the Claude Code session.

One important limitation is that channels only work while the Claude Code session is actively running. For always-on setups, users need to run Claude in a background process or persistent terminal. Events do not queue when the session is closed.

Pros

1. Two-way messaging bridge enables asynchronous AI agent control from any mobile device through familiar messaging apps
2. Full local access means Claude processes requests with filesystem, MCP, and git access against real project files
3. Extensible plugin architecture supports custom channels beyond the initial Telegram and Discord offerings
4. Strong security model with sender allowlists, pairing codes, and enterprise admin controls
5. Webhook receiver capability enables event-driven workflows from CI, error tracking, and deploy pipelines

Limitations

1. Research preview status means the API contract and flag syntax may change, and only Anthropic-approved plugins are allowed
2. Requires active Claude Code session running locally; events do not queue when the session is closed
3. Setup complexity involves creating bots, configuring tokens, and managing pairing codes across platforms
4. Console and API key authentication not supported; requires claude.ai login only

Outlook

Claude Code Channels represents Anthropic's response to a clear market signal: developers want to interact with their AI coding agents through mobile messaging apps, not just terminal windows. The rapid growth of OpenClaw demonstrated this demand, and Anthropic's answer integrates the concept directly into its professional development environment.

The plugin architecture is perhaps more significant than the initial Telegram and Discord support. By publishing source code and providing a channels reference for custom builds, Anthropic is positioning Claude Code Channels as a platform rather than a feature. If the community builds plugins for Slack, Microsoft Teams, and other enterprise communication tools, channels could become the primary interface for remote AI agent management.

The enterprise controls suggest Anthropic is already thinking about corporate adoption. In organizations where developers use Claude Code for production work, the ability to trigger and monitor agent tasks from a mobile device could significantly increase developer productivity during off-hours incidents and collaborative workflows.

Conclusion

Claude Code Channels fills a genuine gap in the AI coding agent workflow by enabling mobile-first interaction with running Claude Code sessions. For developers who need to manage their AI agent while away from their desk, this feature provides a secure, extensible solution. The research preview status means early adopters should expect changes, but the architectural decisions around plugins and security suggest a well-considered foundation for what could become a standard way developers interact with AI agents.