Claude Code Channels launched this week. You can send messages from Telegram or Discord into a running session. Your AI has full filesystem, git, and MCP access. You can code from your phone.
One problem: permission prompts still pause the session. You can't approve them from your phone.
The Gap
You're on a train. Your Claude Code session hits a git push command. It needs approval. The session pauses silently. You don't know until you get home and check the terminal.
Channels gives you input. It doesn't give you approval authority.
Why This Matters for Autonomous Sessions
If you run Claude Code unattended (overnight, during meetings, while you sleep), permission prompts are the #1 cause of silent stalls. Channels doesn't fix this — it was designed for interactive use, not autonomous operation.
The Fix: PreToolUse Hooks
PreToolUse hooks run before the permission system. They can auto-approve safe commands and hard-block dangerous ones — without any human interaction.
{
"permissions": { "allow": ["Bash(*)"] },
"hooks": {
"PreToolUse": [{
"matcher": "Bash",
"hooks": [{ "type": "command", "command": "~/.claude/hooks/destructive-guard.sh" }]
}]
}
}
With this pattern:
- Safe commands: auto-approved, no prompt
- Dangerous commands (rm -rf, force push, git add .env): hard-blocked, exit code 2
- Everything in between: your call
One Command Setup
npx cc-safe-setup
8 hooks. 10 seconds. Covers destructive commands, force push, .env leaks, syntax errors, context window monitoring, and more. 2,354 automated tests.
Channels + Hooks = Best of Both
The ideal setup:
- Hooks handle permission decisions automatically (safe = allow, dangerous = block)
- Channels let you send new instructions from your phone
- No permission prompts = no silent stalls
Channels is great for input. Hooks are great for judgment. Use both.
Is your Claude Code setup actually safe? Run npx cc-health-check — a free 20-point diagnostic. Score below 80? The Claude Code Ops Kit fixes everything in one command.
Top comments (0)