Roadmap
What we're building.
No dates. Items move to Now when they're in active development; they move to the changelog when they ship. Everything on this page is directional — we'll update it when priorities shift rather than leaving stale promises.
Now
In active development. These land with v0.1 or the first public release.
License issuance + Bootstrap URL flow
Buy → email → paste one URL into Claude Code → kit installed. The full credential handshake, device-bound, no long-lived secrets in URLs.
Catastrophe-blocking hooks
Force-push guard, secret-commit scanner, destructive-SQL intercept. Default opt-in; every block has a clean recovery message.
Personal area
License view, billing portal link, project list with last-seen timestamps and recovery actions. Google OAuth.
Project Brain node
Persistent opinionated decisions under .claude/ that Claude no longer has to re-derive each session.
Recovery Readiness dashboard
Live count of which disasters are covered and which aren't, per project. Honest by design — the gaps are shown, not hidden.
Next
Committed to the backlog. No ship date until it's on the Now list.
Second stack support
Expo + Firebase overlay. Same depth as the first stack — named failure modes, cost defaults, fixture-tested hooks. Spike running.
/audit slash command
One command that runs all coverage checks, reports what's drifted, and proposes a repair manifest.
/explain slash command
Ask the kit to explain any installed hook or rule in plain language — what it does, why it's there, how to disable it.
React Flow graph view
The full interactive project graph, replacing the static preview. Pending mobile-graph spike validation at 375 × 667.
Storyblok content integration
Marketing copy managed from CMS instead of hardcoded TSX. Landing page and changelog first.
Later
On the list, but not scheduled. We won't promise dates on these.
/recover slash command
Guided recovery flow for the named catastrophes — step-by-step playbook delivered directly inside Claude Code.
Team licenses
Multiple seats per license, shared project brain across team members. Requires a proper tenant model.
Custom disaster catalog
Let power users define their own named catastrophes with fixture-testable coverage. The public catalog stays public.
Fly.io migration
Move MCP server from Railway to Fly when global latency is an actual user complaint — not before.
Real-time status + alerts
Live monitoring integration for the status page. Ping when the MCP server has a hiccup, not when someone notices it.