Livestream Reply Rule Templates

Set boundaries before automation

The biggest livestream risk is not "slow replies" but misanswers and over-promises in public.

The first version should focus on clear classification, suggested replies, and human override rules—not full auto-reply.

Template 1: Interaction classification

Livestream Interaction Classification

A: Basic inquiry
- Specs
- Schedule
- Activity rules

B: Conversion lead
- Purchase intent
- Package inquiry
- Promo questions

C: High-risk
- Refunds
- Complaints
- Legal / medical / compliance

D: Noise
- Spam
- Off-topic
- Flame / trolling

Template 2: Human override rules

Human Override Rules

Always human-handled:
1. Price commitments
2. Special offers
3. Refunds and post-sale support
4. Legal, medical, compliance
5. Escalated conflict or negativity

Suggest first, human confirms:
1. Activity rules
2. Product specs
3. Basic scheduling

Template 3: Response template

Livestream Reply Template

Question type:
Suggested reply:
Avoid saying:
When to escalate:

Template 4: Per-session review checklist

Livestream Session Review Checklist

[] Frequent questions consistently classified
[] High-risk items all escalated
[] Reply suggestions coherent across session
[] Misanswer hotspots recorded
[] Rules to add/update for next session noted

When not to expand automation

Hold off if:

• High-risk questions are often misclassified
• No clear owner for human takeover
• Review notes are not being kept
• Reply standards vary too much between sessions

Suggested workflow

1. Classify a batch of real comments with the table above
2. Draft suggested replies for A and B only
3. Define escalation triggers for C
4. Run a few test sessions and review
5. Iterate on rules before increasing automation

External links

• OpenClaw docs
• OpenClaw GitHub
• Security checklist
• Troubleshooting