# Manifesto: 12 rules of intelligent communication with AI

AI Workshop · episode 6: Rules of intelligent communication with AI
https://mihalkevich.com/en/workshop/communication

- [ ] Clarity before speed
- [ ] Context is work, not a luxury
- [ ] Define done before you start
- [ ] Plan first, code second
- [ ] Trust, but run it
- [ ] Targeted edits, not rerolls
- [ ] Honesty both ways
- [ ] Polite persistence
- [ ] Least privilege, no secrets in the chat
- [ ] The agent works inside a harness
- [ ] Know when to stop
- [ ] Responsibility stays with the human

## Harness in this episode

The manifesto is the cultural layer of the harness: rules that hold the line before automated checks exist.

## Official documentation

- [Claude prompting best practices](https://platform.claude.com/docs/en/build-with-claude/prompt-engineering/claude-prompting-best-practices)
- [Claude Code — best practices](https://code.claude.com/docs/en/best-practices)
- [Anthropic — Building effective agents](https://www.anthropic.com/engineering/building-effective-agents)

— mihalkevich school