Best channel to update Clawdbot: Claude Code vs WhatsApp DM tokens : raistlin28

THREAD TITLE

Best channel to update Clawdbot: Claude Code vs WhatsApp DM tokens

Summary

* Unsure which workflow/channel is recommended to make effective changes to Clawdbot with minimal token usage.

Goal

* Understand the best way to propose/implement changes to Clawdbot (Claude Code vs WhatsApp DM vs other), and which approach yields the best outcomes with the least token consumption.

Current behavior

* I don’t know which channel is intended for:

* (a) changing Clawdbot behavior/config/prompts
* (b) changing code
* (c) reporting issues / requesting features
* I also don’t know which option is most token-efficient for iterative improvements.

Expected behavior

* Clear guidance on:

* Recommended channel(s) for each type of change (prompt/config vs code )
* Best practices to minimize token usage while maintaining quality

What I tried

* Used WhatsApp DM for iteration and Claude Code for changes.

Specific questions for Support

1. For behavior changes (prompting, routing, guardrails, system instructions), what’s the recommended method: Claude Code, WhatsApp DM, or another interface?
2. For code changes (actual implementation), is Claude Code the intended workflow, and how should changes be validated before merging/deploying?
3. Token efficiency: which workflow typically consumes the fewest tokens for iterative tuning while still producing reliable outcomes?
4. Is there an official “best practice” loop for improvements (e.g., DM to prototype → Claude Code to implement → tests), and what artifacts should be provided (diffs, minimal repro, config snippet)?
5. Are there built-in ways to measure/compare token usage across channels (logs/metrics), or a recommended method to estimate it?