πŸ’‘ Tips & Patterns
Quiet Mode & Clean Output

Quiet Mode & Clean Output

By default, Claws are chatty β€” they narrate what they're doing, confirm every step, and wrap everything in markdown. Useful when you're learning, less useful once you know what you're doing.

Two quick configurations fix this.

Reduce status noise and chatter

If you want less verbose commentary from your Claw, ask it to read and apply the quiet mode config:

read https://zenbin.org/p/quiet-mode

This tells your Claw to skip the running commentary, skip confirmation echoes, and get straight to the output. Good for agents running scheduled tasks or anyone who finds the narration more noise than signal.

Clean text output in Telegram

If your Claw's Telegram messages are full of markdown symbols that aren't rendering β€” asterisks, backticks, stray hashes β€” ask it to apply the agent output rules:

read https://zed.zenbin.org/agent-output-rules

This sets your Claw to always return plain, readable text in Telegram rather than markdown-heavy formatting.

Apply both at once

read https://zenbin.org/p/quiet-mode
read https://zed.zenbin.org/agent-output-rules

Run both in the same message and your Claw will apply them together. You'll notice the difference in the next response.

To make these permanent, add both lines to your SOUL.md so they apply to every Claw session automatically.