Artifact scoring agent
Score artifacts.
Beat the median.
rig-deviant scores any artifact — copy, code, strategy — against the anchor and rupture engines and demands seeded, evidence-rich text clears 60+ on both while generic filler stays under 40. Deterministic, over a fixed corpus, every run.
rig-deviant · anchor + rupture scoring · median-escape engine[higgsfield: gold-lit dashed circle at center with six arrows radiating outward into darkness, one breaking clean past a faint median ring, noir studio]
Why teams install
What rig-deviant earns back
Seeded artifacts clear the floor by 34-40pts
Evidence-rich text scores 100.0 anchor / 94.1 rupture minimum — a 34-40pt margin over the 60 floor.
‹ef33e34›Generic filler can’t fake its way in
Filler artifacts top out at 0.0 on both engines — a 40pt margin under the ceiling, deterministically.
‹42e4471›Runs cold, same corpus, same numbers
5 rich + 5 filler texts, fixed corpus — six consecutive sealed runs produce identical margins.
‹826a956›How it works
Three steps. No faith required.
01
Anchor
Score how tightly the artifact’s claims are bound to real, checkable evidence.
02
Rupture
Score how much the artifact breaks from inherited-orthodoxy phrasing and median AI output.
03
Gate
Seeded artifacts must clear 60 on both; filler must stay under 40 on both, or the packet refutes.
Proof band
One packet. Every claim accounted for.
packet rig/rig-deviant · deviation engine v1 corpus 5 rich + 5 filler · fixed text set · 6 consecutive seals result seeded_min anchor 100.0 · rupture 94.1 (both ≥ 60 floor) result filler_max anchor 0.0 · rupture 0.0 (both < 40 ceiling) anchor PASS — margins hold across 6 consecutive sealed runs ‹ef33e34› themis SEALED — merge gate armed
Every rig-deviant run emits a packet like this — hash-chained, verifier-signed, and queryable from your own dashboard. How ProofPackets work →
Buyers
Who runs rig-deviant
Head of Content
AI-augmented editorial team
Growth Marketer
conversion-copy shop
Solo Founder
landing-page-heavy product
Objections
Answered before you ask
Isn’t “sounds more human” subjective?
Anchor and rupture are scored against a fixed corpus with a hard floor (60) and ceiling (40) — seeded evidence-rich text and generic filler land 34-40 points apart, deterministically, every run.
Can I game the score by just adding more adjectives?
No — rupture specifically penalizes inherited-orthodoxy phrasing and median AI patterns. Filler text tops out at 0.0 regardless of word count (47-55 words across the filler set).
Why trust a score instead of reading the copy myself?
Read it too — but the score is the same 6-consecutive-run, hash-chained number a second reviewer would reproduce. It’s a floor, not a replacement for judgment.
Pricing
Start free. Pay when it pays.
Generalist ships today. Domain Expert adds vertical memory + priors. Custom is built to your stack.
Free
Priced after a short conversation
50 scores a month, anchor engine only.
Team
Priced after a short conversation
Unlimited scores, both engines, corpus history.
Pro
Priced after a short conversation
Dedicated scoring fleet, custom engine weighting, SLA.
First 100 install slots lock Team pricing priced after a short conversation — standard rate is priced after a short conversation.
Runs well with
