Detailed Roadmap

Working name: Sentinel 🛡️ — the unified ops platform vision for monitoring every Coreshift HQ app from one place. Phase 0 product: KeyContent Maintenance Ops (currently the only app on Sentinel).

Last updated: 2026-05-11 Owner: Abe (operator) · Claude Code (implementer) North Star: Catch bugs before users do · Ship fixes in hours · Retire email-support workflow · Scale to every Coreshift HQ app

Current Status

This Week — Pre-CEO Pitch (by Mon, May 18)

Tasks that strengthen the Monday pitch.

Promote V0 from staging to production

• Deploy report-issue Edge Function to the production Supabase project
• Set POSTMARK_SERVER_TOKEN + REPORT_TO_EMAIL as secrets on production Supabase
• Run storage migration on prod — bug-report-screenshots bucket + RLS policies
• Verify Edge Function CORS allowlist includes the production origin
• Confirm Postmark is sending from a live server (not sandbox) with verified sender
• Merge staging branch to main → confirm Cloudflare deploys to production
• End-to-end smoke test on production (Bug + Suggestion + screenshot)

Pitch-strengthening

• Set up inbox rules for [KeyContent · BUG] / [· SUGGESTION] / [· QUESTION] (filter, flag, route)
• Build at least one "Coreshift HQ Way" artifact to attach to the pitch — proof of elevation
• Optional: short dry-run of the pitch (15 min, with timer)

"Coreshift HQ Way" Artifacts (Elevation Layer)

These are the small playbook docs that turn industry standards into your version of those standards. Each one is ~1 page.

• HOW-WE-DO-PRIORITY.md — P0–P3 + user-impact narrative + blast radius
• HOW-WE-DO-PR-REVIEWS.md — non-coder behavior-based PR review checklist
• HOW-WE-DO-BUG-REPORTS.md — the widget + email pipeline + inbox routing (codifies what already shipped)
• HOW-WE-DO-DEPLOYS.md — preview deploy → test → gradual rollout → monitor
• HOW-WE-DO-INCIDENTS.md — alert → diagnose → fix → status page → post-mortem

Recommended order: Priority → PR Reviews → Incidents → Bug Reports → Deploys.

Phase 1 — Full Maintenance Ops Rollout (5 weeks, post-CEO approval)

The 5-week plan from the pitch deck. Week 3 already shipped.

Week 1 — Foundation

• GitHub Issues — labels (bug, enhancement, question, tech-debt, user-reported) + priority labels (P0 → P3)
• Issue templates (bug, feature, question)
• GitHub Project board (Triage → To Do → In Progress → In Review → Done)
• Sentry account + Sentry project for KeyContent
• Sentry SDK installed in the app (frontend + backend)
• Sentry release tracking wired to Cloudflare deploys

Week 2 — Visibility

• Better Stack account + uptime monitor for keycontent.ai
• Public status page at status.keycontent.ai
• Cloudflare alert integration (Workers + Pages health)

Week 3 — User Channel ✅ DONE

• V1 Report Issue widget shipped (May 11, 2026)
• Round 2 polish + ticket-type selector shipped

Week 4 — Workflow

• Briefing templates for Claude Code (one per ticket type)
• PR review checklist published in repo
• Morning triage routine documented (10-min ritual)
• Triage rhythm locked in (when to look at Sentry, Better Stack, GitHub Issues)

Week 5 — Documentation

• PRIORITY.md in the keycontent repo (in-repo summary of priority framework)
• Incident runbook in the repo
• Operator onboarding doc (in case a second operator joins)

Phase 2 — Widget Enhancements (post-rollout, ~3-4 weeks of work)

The widget shipped in Phase 0 grows outputs without a frontend rewrite.

• Add bug_reports table in Supabase — Edge Function writes a row alongside sending the email
• Auto-create a GitHub Issue from each report (mapped: Bug → bug label, Suggestion → enhancement, Question → question)
• Auto-reply email to the user when their issue gets resolved (closes the loop, builds trust)
• In-app "My past reports" view for users (transparency upgrade)
• Rate limiting if abuse becomes a concern (only if needed)

Phase 3 — Scaling Across Coreshift HQ (when launching the next app)

• Document the cloning playbook (PLAYBOOK-cloning-to-new-app.md)
• First clone — apply the system to Coreshift HQ App #2 when it launches
• Federate Sentry org / Better Stack workspace across the portfolio

Phase 4 — Sentinel (the unified ops dashboard, ~when at 2-3 apps)

The single internal platform that aggregates every app's bug reports, alerts, and triage state into one queue. Internal-only — not customer-facing.

Forward-compatible decisions (make NOW, even at one app)

• Add app_id (or app_slug) field to every bug report payload — even though only "keycontent" exists today, the schema is ready for "coreshift-app-2", etc.
• When the bug_reports table is created in Phase 2, include app_id as a column from day one
• Edge Function payload accepts and validates app_id (default to "keycontent" for now)

Build phase (when ready)

• Decide architecture: centralized Sentinel Supabase project OR federated reads from each app's Supabase via API
• Build the Sentinel web app — internal dashboard with triage queue, filters (by app, type, priority), resolution flow
• Auth: Coreshift HQ team only (no public access)
• Optional: federated views into Sentry + Better Stack + GitHub Issues across the portfolio

Operational Maintenance (ongoing)

• Monthly: review issue trends — are the same kinds of bugs recurring?
• Quarterly: review free-tier usage vs limits (Sentry events, Resend emails, etc.)
• Quarterly: prune stale issues, close P3s that have aged out
• Annually: re-evaluate tool choices (Sentry vs alternatives, Better Stack vs alternatives)

Notes / Constraints

• Budget: Free tiers only for new tooling at current scale (~10-20 users). Upgrade triggers documented in deck Slide 11.
• Role split: Abe triages + briefs + verifies. Claude Code implements. Tools watch + alert + record.
• Stack: GitHub (keycontent), Supabase (staging + prod), Cloudflare (Workers/Pages/D1/KV/R2), Postmark (transactional email).
• All artifacts live in: D:\Coreshift HQ\Sentinel\