PublicMITGrok-only

X Mentions auto-reply Bot × SlimeTree-RLM

▸ Auto-reply to @ mentions at a stable quality bar

Fetch @ mentions to yourself → the RLM classifies as D/μ/R → D = instant / μ = silent / R = delegated to Grok. FAQs are handled 0-token; only complex cases hit Grok.

1. X API credentials

User Access Token (scopes tweet.read + tweet.write). Mentions read is available on Free tier; replying requires Basic ($200/mo) or higher.

User Access Token

Your User ID (mentions target)

2. Grok (xAI) API key

Grok key

Platform-native LLM: X mentions processing is Grok-only.

2.5 Operating mode ★ NEW: B (within-Grok tier) ― shared module

Single LLM Cost-tier (B) ― per R mention: cheap → judge → escalate only the insufficient to premium

3. Mentions simulator / batch processing

Paste pseudo-mention text (one per line) and click "Batch process", or fetch via API first and then process.

Processed: 0

D: 0

µ: 0

R: 0

4. Log

(not yet run)

5. Effects

Instant at D: FAQ such as "business hours", "product specs", "pricing" are answered 0-token, 24h
Silent at μ: investment solicitations / spam mentions are not replied to (avoiding pile-on friction)
Grok at R: only borderline cases hit the X-native LLM
WAL chain: records "how the bot judged that mention and what it replied"