Roadmap

v0.1 → v1.0 code-complete in this push. Items marked done include features whose schema, RLS, Edge Function code, and client UI are all shipped to main. Items marked 'blocked' need an external action: API keys, Cornell BirdNET license, ML model training, Apple Developer account, Cloudflare R2 account + DNS, etc. Phases v1.5–v2.5 remain unchanged from prior plan. | 2026-04-25: First real user onboarding session (Eugenio Padilla). 10 bugs found and fixed, 5 new modules designed. Module 14 (User API Tokens) shipped. WebLLM set as default fallback. Identification block added to obs form. | 2026-04-27: v1.1 cross-cutting drops + on-device camera-trap pipeline. Module 22 community validation, owner observation CRUD, atomic delete-observation Edge Function, suggest-from-share-page, OG pipeline (satori build-time + client-side at sync), onboarding v2 (4-step pipeline setup with WCAG focus trap + replay event + telemetry + Anthropic key live verify), MegaDetector v5a YOLOv5 ONNX on-device camera-trap pre-filter (FilteredFrameError short-circuits cascade for empty/human/vehicle frames), bbox-aware species cascade (Phi-vision pre-crops to MegaDetector's bbox), CI smoke check for model-asset URLs, infra/megadetector hosting recipe (convert.sh + INT8 quantise → ~134 MB R2 upload), share/obs locale-neutral 404 fix. | 2026-05-08: Persuasive Tech (Fogg) audit phases 1-2. New phase v1.1.5 captures the principles-driven UX additions: Tier S (Reduction + Tunnelling + Praise + Self-monitoring + Social Facilitation + Responsiveness) shipped via Quick Observation (#759), EXIF-driven photo praise (#757), falta-dex taxonomic gaps (#760), active-observers banner (#758), public triage SLA dashboard (#761). Tier A (Real-World Feel + Reputed Credibility + Transparency) shipped via Field theme for direct sunlight (#768), humanized About page (#770), WhyAmISeeingThis algorithmic disclosure (#772). Cleanup PR #766 removed disambiguation banner / nearby-similar card / places-near-me whose intent is now better served by Fogg replacements. Ola 2b (Conditioning + Kairos + Cause-and-effect + Suggestion + Normative Influence) is in flight: seasonal theme variants (#769), peer-norm bars on license/privacy (#771), tú-vs-promedio percentile card (#773), contextual species suggestions (#774); badge surprises (#727), golden-hour push (#724), ecological-impact viz (#728) remain as open issues. | 2026-05-09: Ola 2b mostly complete. Six PRs merged in a single push — seasonal theme variants (#769), tú-vs-promedio percentile card (#773), contextual species suggestions (#774), mi impacto ecológico 5-metric retrospective (#775), golden-hour kairos push (#777), and variable-reward sorpresas (#779). v1.1.5 phase now reads 14/16 items done; only PR #771 (peer norms on license/privacy) remains in CI. Two cause-effect / kairos / variable-reward items that had been parked as open issues (#727 / #724 / #728) shipped honestly inside the same window — fixed catalogs, hard daily caps, opt-in default-OFF, rules disclosed in /docs/surprises and /profile/notifications.