Guide (English)

• Project checkpoint, 6 November 2026

Project documentation in English.

Engineering leadership

• Production readiness and SRE - launch gates, security, recovery, GDPR jobs, automation and incident response; evidence is not yet recorded.

• Privacy, trackers and scan statistics - legal documents, consent, private valuation and aggregate counters for website/mobile/desktop.

• CTO playbook — audit reconciliation, roadmap, team guardrails, security/GDPR, load capacity, and the five workflow variants by budget. Start here before any team or release work.

What shipped

• Feature and fix tracking — every feature and fix, newest first, generated from one file per entry. The format and the reasoning are in tracking/README.md.

Automation

• The agent runner — how the VPS picks up agent:ready issues, implements them, reviews them, and stops short of merging: the label flow, one cycle step by step, which skills each repository gives the agent, the cron schedule, the kill switch, and what actually protects you

Infrastructure

• Administration console: architecture, public access status, responsibilities and decisions.

• Admin data and features: catalogue metadata, media workflow, Benchmark and support.

• Admin security and accounts: staff separation, roles, passwords, sessions and deferred MFA.

• Admin operations: protected backups, migrations, releases, ingress reload, rollback and troubleshooting.

• Container inventory (PC) — what runs locally on the Windows host today

• Docker PC → VPS migration — how to bring the stack up on the VPS

• VPS deployment over Tailscale — concrete plan: replace the laptop, keep the phone on the tailnet, serve images from our nginx

• Public VPS deployment (domain + HTTPS) — production target: Caddy as reverse proxy with automatic TLS, Tailscale out of the application path

• VPS / deployment troubleshooting — recurring pitfalls (docker group, whiptail, gpg, RTK, gh auth) and their fixes

Mobile platforms

• Account deletion - confirmed in-app erasure, API contract, device cleanup, destructive QA and retention boundaries.

• Apple build scan privacy and retention - verified scan behaviour, proposed retention, enforcement evidence and release boundaries.

• iOS readiness — there is no ios/ project at all; which blockers are real (Apple licence, no camera on the Simulator) and which are already solved (macOS builds), plus why Expo does not change the answer

Sessions

• 2026-05-05 — Option B deployment end-to-end — full story: Tailscale, nginx mountpoint fix, SSH key-only, image scan debug (cleartext, .env not loaded, FAISS coverage)

Org / infra changes

• GitHub org renames — tcg-alpha → skn-tcg → snag-tcg — both renames (2026-05-13 and 2026-07-30), what moved, what deliberately did not, and how to update local clones

Product strategy

• Pokecardex rarity reference: complete collaborator-supplied benchmark, source date and ordered rarity labels for 130 series.

• Product engine architecture — product map for identity, collection, pricing, and future marketplace engines

• M1, M2, and M3 valuation engines — raw, sealed, and graded rules, implementation status, and closed marketplace API posture

• Persona and positioning — sealed-investor persona, product edge, and demo narrative

• Competitive landscape — direct competitors, high-threat scanner apps, and monitoring criteria

• Data sources and pricing strategy — source layers, provider posture, and explainable pricing rules

• Collection, pricing, and alerts backend plan — backend domains, tables, APIs, and phased roadmap for portfolio valuation and market alerts

• Asset licensing and source risk register — operating rules for internal assets, public serving, provider terms, and production-safe data

• PRD feature map — MVP feature groups and later roadmap boundaries

• Complete mobile UI/UX feedback register — consolidated Nini/Lulu requests, product decisions, delivery status, and remaining dependencies

Scan & TCG integration

• One Piece classification rules — exact physical distribution determines series membership; printed gameplay numbers group all editions.

• Pokemon rarity business rules - reviewed series order, EX/SWSH groups, historical secrets and compatibility

• Live Pokecardex order comparison - all source profiles, series mappings and ordering-only proof

• Scan resources — external scanner references, community prototypes, and API/data links to review before use

• Scan evaluation protocol — repeatable metrics for scanner latency, endpoint accuracy, phone captures, and correction success

• Scan algorithm inventory — current Lab, Compat, Tormknd, Combined, Frame, Sealed, Booster Lab, and POC fixture roles

• Scan UI/UX competitive review — camera-first scanner UX direction from MonPrice, TCG CardScanner, Iestims, and Zebradex references

• Sealed Explorer catalogue policy - shared series tree, packaging categories, chronology and item details

• Sealed scanner lab — backend lab path for sealed products using raw DINOv2 embeddings, pgvector, and internal-only ptcg-assets

• Sealed field photo collection format — photo protocol, CSV columns, rights rules, and batch handoff format for associates

• Sealed photo instructions for associates — simple non-technical checklist for photos usable by YOLO, DINOv2, and scanner evaluation

• Product & marketing brief — improving sealed detection — what is missing to move from the sealed lab to reliable real-world detection

• Recognition optimization strategy — working direction for DINOv2/FAISS, YOLO, OCR, reranking, sealed/slab recognition, and benchmark-driven improvements