CareConnect: Product Roadmap
Current Version: v22.0 (Non-Duplicate Value Decision Plan, Phase 0) Next Milestone: v22.0 Gate 0 Exit (C1/D4 blocker closure) Last Updated: 2026-04-05 Platform Status: Strategic Repositioning - v22.0 Decision-Gated Planning
Current State
- Services: 196 manually curated social services (
npm run validate-dataandnpm run audit:dataon 2026-04-03) - Tests: default Vitest suite green as of 2026-04-03 (
165files;1251passed;24skipped) - DB integration lane: local Supabase-backed retrieval, route, export, search, and policy tests are green via
npm run test:db - Coverage:
72.13%statements /78.85%branches /83.20%functions /72.13%lines fromnpm run test:coverageon 2026-04-03 - Repo hygiene:
npm run check:refs, typed service DB write paths, dashboard server actions, and dependency cleanup are complete - Dependency audit:
npm audit --omit=devreports0 vulnerabilities - Bundle baseline: localized home route first-load JS is
315 kBafter lazy AI and semantic-search startup deferral - E2E: default Chromium suite is skip-free; production/server-mode checks live in dedicated opt-in commands
- Accessibility: WCAG 2.1 AA automation remains in place
- Languages: 7 locales at translation-key parity
- Dashboard trust/resilience: partner overview metrics are live, degraded read states are explicit, and localized dashboard/admin redirects now preserve locale
- Search explainability: public search results now surface deduplicated match reasons with an accessible disclosure for deeper scoring context
- Stale-data governance: search now excludes records beyond the 180-day freshness window, freshness badges distinguish expired records, and direct-linked detail pages show an explicit stale-record warning
- Freshness policy alignment: governance and planning docs now treat 180 days as the hard visibility limit, 90 days as a priority-service target, and the v22 90-day window as a review checkpoint rather than a guaranteed build schedule
- Workflow runtime hygiene: active GitHub Actions use Node-24-compatible major versions, and release automation uses
gh release createinstead of the archived release action - Pilot metric stack: M2/M4/M5/M6/M7 source schema, recompute path, and scorecard snapshot flow are implemented; values remain data-dependent rather than schema-blocked
- Pilot readiness reporting: scoped JSON/Markdown/CSV readiness exports now exist for bounded A6/A16 follow-through without mutating curated service data
- French service-data gaps:
access_script_fr,hours_text_fr,eligibility_notes_fr, andsynthetic_queries_frremain incomplete - Offline: PWA with IndexedDB fallback, background sync, and snapshot-age/stale-data messaging on offline surfaces
- Privacy-safe mapping: service-detail pages gate third-party map previews behind explicit user action
- Observability: Axiom metrics, Slack alerting, SLO monitoring, and runbooks are live
- Deployment: Live on the direct-VPS path at
https://careconnect.ing, withhelpbridge.caandwww.helpbridge.caredirecting to the canonical host - Branding: CareConnect rename is complete across this repo, the
jerdaw/careconnectGitHub repo slug,platform-ops, and the live VPS runtime - 211 sync posture: quarantined to explicit manual runs only; no scheduled or mock-data ingestion path remains active
- Data quality gaps:
- Scope: 0 missing
- Coordinates (any): 58 missing
- Coordinates (required): 18 missing
- Kingston missing address: 17
- Access scripts: 0 missing
- Structured hours (active services): 10 missing
- Hours text (active services): 10 missing
Decision Summary
CareConnect is not currently in a breadth-expansion phase.
The active question is whether the project can prove non-duplicate value relative to 211 through measurable last-mile outcomes such as connection success, reliability, and referral completion. Until that is demonstrated, roadmap priority stays on governance closure, pilot readiness, and evidence discipline rather than new feature expansion.
What To Do Now
- Close the remaining v22.0 Gate 0 blockers in strict order: C1 legal review, then D4 partner operations evidence.
- Keep the repo stable while Gate 0 is blocked: maintain tests, keep docs aligned, and avoid speculative feature work.
- If pulling forward any admissions backlog work, treat
A3,A11, andA22as complete; onlyA1and boundedA6/A16remain valid near-term Tier 0 execution. - Preserve launch readiness materials, but do not resume beta or public-launch execution until v22 permits it.
What Not To Do Now
- Do not expand directory breadth to compete with 211.
- Do not start new pilot-facing features before Gate 0 evidence is accepted.
- Do not restart v19 launch execution while v22 remains
NO-GO. - Do not pull forward Tier 1-4 admissions packaging work before real pilot evidence exists.
- Do not pull forward parked enrichment or portfolio work unless it directly supports the active gate.
Active Work
v22.0: Non-Duplicate Value Decision Plan 馃攧 ACTIVE
Status: Phase 0 in progress - Gate 0 Exit NO-GO Priority: Critical Timeline: target 90-day decision review cycle (~13 weeks), contingent on external Gate 0 dependency closure Created: 2026-02-27
CareConnect is being repositioned from possible directory duplication toward measurable last-mile outcome value. This path uses explicit hypotheses, strict kill criteria, and governance gates before any broader pilot execution.
The 90-day window is a review target rather than a guaranteed engineering schedule. It is only realistic if legal/API review and partner-operations evidence arrive early enough to support a real decision inside that window.
Core objective
- Prove non-duplicate value vs 211 on measurable connection outcomes.
- Validate privacy-safe integration feasibility with 211 pathways.
- Keep explicit stop conditions if outcome value is not demonstrated.
Current technical position
- Pilot DB schema, RLS policies, internal pilot APIs, and pilot tests are implemented.
- Step 1 approval locks are complete.
- Gate 0 evidence scaffolding is in repo.
- C2 retention policy approval, privacy sign-off, and dated verification evidence are complete.
- Gate 0 remains
NO-GObecause C1 legal evidence and D4 partner-ops evidence are still incomplete.
Immediate blockers
G0-3 / C1: candidate partner legal/API terms are not yet attached for clause-level review.G0-8 / D4: named pilot partner list, outreach ownership, and dated execution evidence are missing.- Baseline execution is recorded, but current M1/M3 values are
NULLbecause the baseline window had zero events.
Required user-owned actions
UA-1: attach candidate partner legal/API terms and complete clause-level C1 review.UA-3: attach named pilot partner list, outreach owner assignment, and dated outreach execution evidence.
Agent follow-through once evidence exists
- Update the Gate 0 trackers and source control docs.
- Sync the integration decision record, approval checklist references, and evidence matrix.
- Re-evaluate Gate 0 and re-run
npm run check:v22-gate0. - Keep pilot APIs, schemas, tests, and docs aligned with any approved control changes.
Gate 1 success thresholds
- Failed contact attempts reduced by at least 30% vs baseline
- Time-to-successful-connection reduced by at least 25%
- Freshness SLA compliance at least 70%
- Referral outcome capture at least 50%
- Fatal data-decay error rate at or below 10%
Canonical references
- v22.0 Non-Duplicate Value Decision Plan
- v22.0 Approval Checklist
- v22.0 Phase 0 Implementation Plan
- v22.0 Gate 0 User Action Tracker
- v22.0 Gate 0 Evidence Intake Pack
- v22.0 Gate 0 Exit Checklist
Parallel Maintenance While Gate 0 Waits
These items are worth doing only if they do not distract from Gate 0 closure:
- Keep the default E2E suite skip-free and keep the opt-in production/server suites healthy.
- Verify and document the remaining v22 threat-model mitigation items before pilot activation.
- Run
npm run db:typeson a Docker-capable machine and remove the last intentional untyped admin-audit access once generated schema coverage exists. - Expand exact-English duplicate i18n auditing from the current focused namespaces to all used translation keys after the remaining legacy translation debt is localized.
- From the admissions backlog, only execute the still-open Tier 0 items that strengthen pilot readiness or evidence discipline:
A1and boundedA6/A16.
On Hold
v19.0: Launch Preparation 鈴革笍 ON HOLD
Status: Automation and documentation complete; human execution deferred pending v22 Gate 0 Priority: High, but subordinate to v22
The launch-prep stream is intentionally paused. The repo-local automation is already built, but the remaining work is manual QA, beta execution, and launch operations. Resume only after v22 permits it.
When resumed
- Run the production environment audit.
- Execute critical user-journey testing.
- Complete the top-20 service review.
- Resume beta operations and launch monitoring.
References
- v19.0 Launch Preparation Plan
- v19.0 User Execution Guide
- v19.0 Phase 1 Execution Handoff (2026-03-09)
v21.0: Admissions Portfolio & Launch Narrative 鈴革笍 PARKED
Status: Parked until after v22 Gate 0 / Gate 1 evidence, but fully re-triaged on 2026-04-01
This backlog is now explicitly sorted by admissions value under the current v22 constraints. The sequencing rule is:
- Close v22 blockers first.
- Then create real pilot evidence.
- Then package that evidence for admissions and external audiences.
Allowed now only if they directly support v22
- Close Gate 0 partner/legal evidence (
UA-1,UA-3). - Run bounded verification and pilot-scope data-quality work when it improves pilot readiness.
- Preserve the completed Tier 0 hardening bundle as baseline capability, not as a reason to pull forward packaging work.
First work after Gate 0 exit
- Run a small real-world pilot in one actual referral workflow.
- Formalize the 211 boundary/handoff, run crisis-safety validation, add a quick-exit path for violence-sensitive browsing, and conduct professional usability sessions.
- Build partner/referrer collateral, expose logistics needed for referrals, add shortlist/multi-print referrer workflows where pilot users need them, begin L3 outreach, and recruit the first advisory reviewers.
First proof artifacts after live pilot activity starts
- Publish a baseline-to-pilot scorecard and decision memo.
- Secure factual support letters and first L3 confirmations.
- Publish transparency and status surfaces, plus drill/accessibility/equity evidence.
Later packaging work
- External privacy/AI review.
- Leadership/collaboration visibility updates.
- Presentations, poster/case-study, and other dissemination artifacts.
This work remains strategically useful, but it depends on real operational evidence. Keep it parked until the v22 pilot produces something defensible to package.
Reference:
v20.0: Testing and Technical Excellence 鈴革笍 MAINTENANCE MODE
Status: Most autonomous backlog work is complete; only bounded maintenance remains active
Useful maintenance items:
- Keep the default E2E suite healthy.
- Keep the dedicated DB integration lane healthy and deterministic.
- Publish a fresh coverage snapshot.
- Fix repo-local regressions surfaced by routine validation.
Deferred items:
- Advanced French service-data enrichment
- Search AI metadata migration out of JSON
- Admin-facing data quality dashboard
- Regenerate
types/supabase.tsusingnpm run db:typeson a Docker-capable machine and then type the remainingnotification_auditpath - Unify client and server ranking around one shared scoring engine and retire the remaining placeholder scoring path
References:
- 2026-02-12 v20.0 Phase 1 Implementation Plan
- 2026-03-12 v20.0 Autonomous Backlog Closeout
- 2026-03-24 v20.0 DB Integration Test Lane
- 2026-03-24 v20.0 Supabase Migration Recovery
Completed Work
Recent Completed Milestones
- Gate 0 wait maintenance bundle (2026-04-05): completed search explainability, stale-data runtime governance, workflow-runtime doc alignment, and solo-scale freshness-policy calibration while Gate 0 remained blocked; archived in 2026-04-05 v22.0 Gate 0 Wait Maintenance Bundle.
- Map privacy and offline snapshot safety surfaces (2026-04-04): replaced automatic Google Maps embeds with explicit opt-in previews, surfaced offline snapshot age/stale warnings on offline surfaces, updated threat-model/user-guide/architecture docs, and added focused UI/helper coverage; archived in 2026-04-04 v22.0 Map Privacy and Offline Snapshot Safety.
- Tier 0 admissions-support hardening (2026-04-01): completed A3 pilot metric instrumentation, A11 public-claim hardening, A22 focused pilot/privacy test coverage, and bounded A6/A16 readiness-audit tooling; archived in 2026-04-01 v22.0 Pilot Metric Instrumentation and Tier 0 Hardening.
- CareConnect production cutover and repo finalization (2026-04-03): completed the live
careconnect.ingVPS cutover, legacy HelpBridge-domain redirects, GitHub repo rename tojerdaw/careconnect, and post-cutover observability-noise cleanup; archived in CareConnect Rebrand Archive. - C2 retention control closure (2026-03-29): approved retention policy, captured privacy sign-off, attached dated read-only verification evidence, and moved
G0-4topass. - Workflow/runtime cleanup and 211 sync quarantine (2026-04-01): upgraded the remaining Node-runtime-sensitive GitHub Actions, replaced archived release creation with
gh, removed placeholder 211 sync records, and restricted the 211 sync path to explicit manual execution only. - Audit remediation hardening (2026-03-30): enforced org-scoped service creation, removed fabricated provenance, fixed Slack/runbook links and dashboard CSV parsing, added focused runtime coverage, repaired noisy scheduled workflows, and reduced the localized home-route first-load JS to
315 kB. - Repo audit remediation (2026-03-29): completed the typed service-write cleanup, feedback/dashboard action consolidation, member-management split, privacy-safe analytics hardening, reference validation, and dependency/script hygiene follow-through.
- Dashboard trust, resilience, and locale hardening (2026-03-30): replaced placeholder partner metrics with live 30-day summaries, added deterministic degraded states for impact/analytics/feedback pages, localized observability/admin surfaces, and tightened focused i18n duplicate-English enforcement.
- Code quality remediation (2026-03-26): 0 npm audit vulnerabilities, console鈫抣ogger migration, component reorganization, schemas rename, ESLint strictness for lib/, file cleanup.
- Premium Dark Mode (2026-03-25): Redesigned dark theme with high-contrast Slate-950 base, pure white text, and structured elevation.
- v20.0 migration recovery (2026-03-18): 41-file migration chain collapsed into a single reproducible baseline + 3 forward migrations, test infrastructure unified on migration-linked bootstrap.
- v20.0 maintenance (2026-03-18): real DB-backed Supabase retrieval/policy test lane added with blocking CI coverage
- v18.0: Production observability, Slack alerting, SLO tracking, observability dashboard, and runbooks
- v17.7: Search quality testing framework and scoring refinements
- v17.6: Authorization resilience, translation workflow automation, and load-testing baseline
- v17.5: Circuit breaker rollout, performance tracking, health/metrics endpoints, and k6 infrastructure
- v17.0-v17.4: Security, test coverage, internationalization, accessibility, and partner portal foundations
Why This Matters
The project already has the technical base for a live, privacy-first, resilient service directory. The open question is no longer whether the app can ship technically. The open question is whether it creates enough non-duplicate operational value to justify continued expansion.
Archive and Historical Plans
- Planning Archive
- v22.0 Gate 0 Wait Maintenance Bundle Archive
- v22.0 Map Privacy and Offline Snapshot Safety Archive
- v22.0 Pilot Metric Instrumentation and Tier 0 Hardening Archive
- v20.0 Runtime Hardening and Performance Remediation Archive
- v20.0 Workflow Runtime Cleanup and 211 Sync Quarantine Archive
- v20.0 Repo Audit Remediation Archive
- Code Quality Remediation Archive
- CareConnect Rebrand Archive
- v20.0 DB Integration Test Lane Archive
- v17.6 Archive
- v17.5 Archive
- v17.4 Archive
- v17.3 Archive
- v17.2 Archive
- v17.1 Archive
- v17.0 Archive
Review Triggers
Update this roadmap when any of the following happen:
UA-1,UA-2, orUA-3changes status.- Gate 0 moves from
NO-GOtoGO, or is re-affirmed asNO-GO. - v19 resumes or is explicitly deferred further.
- A major data-quality, testing, or deployment baseline changes.
- v21 tier sequencing changes or a new admissions evidence artifact materially changes what is worth prioritizing.
- A new strategic version becomes active.
Operating Rule
If there is tension between adding more capability and closing the current decision gate, close the decision gate first.