Open protocol

MILA · X-RADAR

Ask your social graph questions. It answers with evidence, or it says it doesn't have any.

An intelligence layer for the X social graph. Natural-language control. Evidence, not vibes. Built as a Grok-native operator console.

Vocabulary

• Fingerprint — signed, versioned behavioral object. Schema fingerprint/v2.1. Never contains a field called intelligence.
• Lens — named, composable, shareable search with creator key, timestamp, and hash.
• Radar Sweep — a pass over the reachable universe, not all of X.
• Blind Spot — a hole in your feed: you care about it, you follow nobody who posts it.
• Heated cluster — a group of people yelling about the same thing in a short window. Football Twitter at 11pm. A song jumping into meme accounts.
• Hotspot — that cluster pinned to a city. Dallas at 11:02pm. London an hour later. The map is where and when, not just what.
• Credibility Ledger — public, falsifiable, dated predictions.

Honest constraints

• This console indexes a defined reachable universe (115 accounts in the current snapshot) and expands from operator graphs. It does not search all of X. Live X Search is an explicit, capped probe of public posts.
• Follow/unfollow is proposed here and executed only through authenticated user actions.
• Only public data. No DMs. No private engagement.
• Fingerprints are derived behavioral data. Opt-out and deletion on request.

Attribution

Lens Protocol schema is published for attribution-required reuse. Derived Lens implementations must credit the X-Radar protocol. Ranking engines are not open-sourced. Signed manifests travel with every object: schema version, computing key, timestamp, hash chain.

computed_by xradar/1.0 (key: mila-026-inkline)
publisher MILA / 026 Inkline Labs LLC

Copyright

© 2026 Mila · 026 Inkline Labs LLC. All rights reserved.

Mila, X-Radar, Fingerprint, Lens, Radar Sweep, Blind Spot, and Credibility Ledger are used as product vocabulary by 026 Inkline Labs LLC. Spec v2.0, September 2026.