- Monorepo setup (pnpm workspaces)
- Shared types contract
- Data collection tools (GitHub, DeFi Llama, CoinGecko, Helius)
- Curated Solana repo list (65+ verified repos)
- Z-score anomaly detection
- OpenRouter GLM-4.7 client
- Narrative clustering via LLM
- Build idea generation
- Cross-layer scoring
- JSON + Markdown report output
- Pipeline orchestration (index.ts)
- Website: landing page, report detail, archive, methodology
- CI workflow (typecheck + tests)
- GitHub Actions cron (fortnightly)
- GitLab mirror workflow
- GitHub repo (RECTOR-LABS/solis)
- Configure repository secrets
- First dry run: validate pipeline end-to-end
- First real report generated (2026-02-12)
- Deploy website to VPS (solis.rectorspace.com)
- Docker + GHCR CI/CD pipeline (auto-deploy on push to main)
- Layer 0: X/Twitter KOL timeline signal collection
- Layer 0: LunarCrush social signal collection
- Narrative history tracking (fuzzy matching, stage transitions, report diffing)
- Post-pipeline alerting (Telegram/Discord notifications)
- Pipeline error resilience (retry logic, graceful degradation)
- Dynamic repo discovery via GitHub Search API
- x402 micropayment middleware (opt-in paid tier bypass)
- Report comparison page (/compare)
- Narrative timeline page (/narrative/[slug]/timeline)
- Full-text search modal (Cmd+K)
- CSV/JSON export
- Homepage revamp: 7-section glassmorphic landing page
- Report page revamp: 8-section intelligence dashboard (header, nav, metrics, diff, stage-grouped narratives, filtered ideas, source cards)
- Expand curated repo list to 100+ (103 repos)
- Persistent VPS heartbeat daemon (replaces GitHub Actions cron)
- Docker volume mount for reports (no rebuild on new reports)
- Manual quality review of first 2-3 reports
- Monitor heartbeat daemon reliability (first week)
-
Set up Telegram alerts for pipeline notifications(deferred — low priority, daemon logs suffice for now)
- Historical reports API (x402-gated)
- Signals API (x402-gated)
- Wallet setup (Solana USDC)
- Pricing page / documentation
- Payment flow end-to-end testing
- Side-by-side eval: LLM comparison CLI (
pnpm eval:llm) - Filesystem signal cache (20h TTL, delta-only fetching merged into cache)
- GitHub star history tracking (7d/30d deltas, velocity)
- Confidence calibration analysis (
pnpm eval:calibration)
- Interactive charts (Recharts) — confidence distribution, timeline, radar, meta-trends
- Trends page (/trends) — pipeline stats across all reports
- Custom LLM query API (x402-gated, $0.05/call)
- Force-directed knowledge graph (/brain) — replaces static brain.html
- Email subscriber management (JSON store, HMAC unsubscribe)
- Email digest via Resend (daily trigger API, HTML template)
- RSS feed for reports
- 2026-03: 1st Place — Superteam Earn "Narrative Detection & Idea Generation" bounty ($2,000 USDG, 117 submissions)
- 2026-02: First live report generated, VPS deployment, heartbeat daemon operational
- Community-contributed repo tracking
- Multi-chain expansion (Ethereum L2s, Sui)
- Prediction market integration (Polymarket)
- API SDK for programmatic access