AI-powered Pokémon TCG centering analysis

Implementation OpenClaw instance Skills Architecture Roadmap

Primary measurement

L/R + T/B

Front and optional reverse centering ratios with uncertainty.

Core grader profiles

PSA CGC BGS

SJC/SGC held as configurable until exact target is confirmed.

OpenClaw additions

4 skills

Capture, vision, grading rules, and report generation.

MVP target

±1–2%

On clean qualifying photos after perspective correction.

What this mini-site defines

The project has moved from concept brief to implementation dossier.

TCGMarketPulse Centering Intelligence should be implemented as a measured eligibility engine: image intake, quality scoring, card detection, perspective correction, layout-aware measurement, grader-rule comparison, and an explainable report.

The OpenClaw instance becomes the orchestration layer for early development and operator workflows: receiving photos, calling vision workers, invoking skills, storing analysis artifacts, and letting PulseLine explain results conversationally.

Page map

Grader tolerance matrix

Initial centering-only profiles for the rules engine.

Profile	Level	Front	Back	Implementation note
PSA	Gem Mint 10	≈55/45	≈75/25	Return “centering eligible,” never a guaranteed PSA 10.
CGC	Pristine 10	50/50	50/50 expected	Very strict result requiring low uncertainty.
CGC	Gem Mint 10	≈55/45	≈75/25	Separate from Pristine tier.
BGS	Pristine 10	50/50	60/40+	Centering subgrade candidate only.
BGS	Gem Mint 9.5	50/50 one axis; 55/45 other	60/40+	Requires axis-specific logic.
SJC / SGC-compatible	Configurable	TBD	TBD	Do not hard-code until exact grader is confirmed.