Public Development Roadmap
A public view of Starnum's product and governance direction. Roadmap entries are plans, not delivery promises.2026-04-12
Release authority remains with the human operator. Machine-readable checks and independent AI verification support the decision; they do not replace it. See the governance model .
Completed Foundations (as of 2026-04-12)
Completed
Platform Infrastructure
✓ iztro open-source chart-engine integration (14 major stars and Four Transformations) ✓ 10-language public-site framework (zh-TW / zh-CN / en / ja / ko / es / vi / th / ms / id) ✓ Cloudflare Workers and static-asset deployment ✓ Cloudflare R2 and D1 chart storage with lifecycle controls ✓ Content integrity and release verification ✓ Machine-readable quality gates across six dimensions
Knowledge Base
✓ Structured Zi Wei Dou Shu and numerology source corpus ✓ Lu Binzhao (Zhongzhou) school as the declared Zi Wei Dou Shu baseline ✓ Knowledge-graph and semantic-retrieval layers ✓ Machine-enforced astrology logic rules ✓ Three-layer evidence retrieval: semantic search, graph expansion, and bounded source slices
Content Operations
✓ Traditional Chinese source corpus with localized article variants ✓ Language-specific translation memory and quality gates ✓ Editorial-standard modules and bounded agent workflows ✓ Sanitized case references for quality comparison ✓ Tier-specific chart-reading assembly and verification
Trust and Transparency
✓ Localized methodology and governance pages ✓ System card, public benchmark, and transparency reports ✓ Security disclosure, acceptable-use, ethics, and accessibility policies ✓ Machine-readable public-evidence manifests and verification commands
Q2 2026 — Phase A: Strengthen Foundations
Target window: April–June 2026
Content Quality
○ Expand the Traditional Chinese source corpus ○ Complete the 2026 annual-forecast source set ○ Repair content flagged by the quality registry ○ Expand sanitized quality-comparison cases
Technical Quality
○ Accessibility audits and prioritized WCAG repairs ○ Core Web Vitals monitoring and regression prevention ○ Maintain standards-based AI discovery surfaces ○ Keep public accessibility evidence current ○ Refresh semantic-retrieval indexes when source fingerprints change
Knowledge Base
○ Complete reviewed source distillation with provenance ○ Expand source coverage without mixing incompatible schools ○ Expand graph coverage while preserving source lineage
Q3 2026 — Phase B: Localization Quality
Target window: July–September 2026
○ Grow each language from the Traditional Chinese semantic source ○ Run language-specific review with one accountable owner per locale ○ Reduce untranslated residue, extraction drift, and unnatural machine phrasing ○ Keep FAQ and governance surfaces semantically aligned by language ○ Document cultural adaptation where literal translation is insufficient
Community and Research
○ Publish research notes on school differences, methods, and responsible AI ○ Evaluate a consent-based browser for sanitized case studies ○ Evaluate privacy-preserving feedback on chart-reading usefulness
Q4 2026 — Phase C: Carefully Bounded Features
Target window: October–December 2026
Candidate Features
○ Compatibility analysis, subject to engine support and responsible-use review ○ Opt-in annual-cycle reminders with clear consent controls ○ Evaluate bounded messaging integrations without open-ended prediction claims ○ Improve mobile chart visualization and accessible summaries
AI and Quality
○ Revision-addressed regression tests for content and chart workflows ○ Quarterly strategy and governance review ○ Expand the knowledge graph only with provenance and verification
2027 and Beyond — Direction
Long-term direction
◎ Build a multilingual Zi Wei Dou Shu and numerology learning platform ◎ Shorten verified chart-reading turnaround without weakening quality gates ◎ Evaluate practitioner-contributed cases with consent, provenance, and review ◎ Explore research partnerships with bounded knowledge-base access ◎ Explore accessible audio summaries after text quality is stable
Explicitly Excluded
The following uses are outside the product roadmap:
✕ Death prediction — prohibited by the ethics policy✕ Medical diagnosis — requires licensed medical professionals✕ Criminality or risk profiling — discriminatory and prohibited by the acceptable-use policy✕ Deterministic decisions in legal, financial, employment, or medical contexts ✕ Public-figure chart analysis without consent — unacceptable privacy and reputational risk
Feedback and Proposals
To suggest a feature or report a problem:
A proposal enters the roadmap only after scope, responsible-use, privacy, and verification review. See the governance model .