Roadmap
What we're building, in what order, and why. Vote on the ones that matter to you — votes shift the priority queue. Got an idea that isn't here? Submit it.
No ideas match . Try a different keyword.
Planned · 9
- 48 Planned ETA · June–July 2026
MCP Server Directory — search, filter, install
Curated, taggable directory of Model Context Protocol servers. Filter by runtime, auth type, popularity. One-click install snippets for Claude Desktop, Cursor, Zed. The second pillar of Prismix beyond the status hub.
- 29 Planned ETA · June–July 2026
Bundles — group MCP servers into stacks
"React dev kit", "research workflow", "data ops" — bundle related servers so others can install your whole workflow in one command. Fork bundles, remix them, share via URL. Private bundles are a Pro feature.
- 24 Planned ETA · August 2026
Manual subscription tracker
Track every recurring AI bill — Claude Pro, ChatGPT Plus, Cursor, Replit, etc. Renewal reminders, currency conversion, CSV export. The first pillar of the personal AI dashboard.
- 23 Planned ETA · July 2026
Reviews + ratings on servers
5-star rating + markdown reviews from people who actually shipped with the tool. Reviewer karma based on bundle adoption.
- 19 Planned ETA · July 2026
Submit a new MCP server
PR-style submission flow — paste a GitHub URL, fill out a short form, we curate. Pro accounts get priority moderation.
- 17 Planned ETA · August 2026
Public profile pages
Each user gets `/u/<username>` — bio, bundles, reviews, follow button. Custom theme colour for Pro accounts.
- 14 Planned ETA · September 2026
Connected providers — pull live API spend
Optional: paste an OpenRouter / OpenAI / Anthropic admin key (AES-GCM-encrypted, AAD-bound to your userId) and we pull live token usage + spend into the dashboard. Provider keys never leave KV unencrypted. Pro feature.
- 14 Planned ETA · August 2026
Personalized feed of AI tooling
Activity from people you follow + global trending. New servers, fresh bundles, headline AI news, status incidents on your starred services. Sorted by relevance, not recency.
- 9 Planned ETA · August 2026
CSV + JSON export of everything
One-click export. Bundles, reviews, saved servers, manual subs. No vendor lock-in — your data, your file.
Ideas · 6
- 12 Idea
Follow developers whose stack you trust
See what they bundle, what they review, what they ship with. The asymmetric "follow" model (Twitter-style, not Facebook-style).
- 9 Idea
Browser extension for Claude.ai / ChatGPT / Cursor
Read your already-visible usage limits (no cookie access), surface them in Prismix. Pair with a per-device token, revocable any time.
- 7 Idea
Public read API (versioned, rate-limited)
GET /api/v1/servers — paginated, filterable. SDK comes after. Build your own MCP launcher. (The /api/v1/statuses + /api/v1/incidents endpoints for the status hub already follow this pattern.)
- 6 Idea
Anomaly detection on API spend + incident frequency
Statistical baselines per provider — flag when your daily OpenAI spend is 3σ above last 30 days, or when Anthropic incident count spikes for the week. Email + webhook alert. Pro feature.
- 4 Idea
Affiliate partnerships with Claude Desktop, Cursor, Zed
"Install via Claude Desktop" buttons that pass attribution. Revenue share goes to maintaining curation.
- 3 Idea
Migrate hot tables KV → D1
KV is great for write-once-read-many key/value. When the directory + feed cross ~10k items, the trending sort + filtered queries cost too many round-trips. Migrate `mcp:server:*`, `feed:item:*`, and `roadmap:item:*` to D1 while keeping KV for caches + sessions.
Shipped · 18
- 67 Shipped ETA · Shipped May 2026
Per-service alert preferences
Tune severity threshold per starred service or mute it entirely, on top of the global default. "Page me on Anthropic minor degradation, only critical on OpenAI, mute Cursor". Pro feature, surfaced inline on /alerts under "Per-service overrides".
- 53 Shipped ETA · Shipped May 2026
Real-time status dashboard for AI services
Aggregated live status of 48 AI providers — model APIs, dev tools, infrastructure, consumer apps. Colour-coded indicators, incident counts, 30-day uptime, cross-service incident feed, star your favourites, per-service detail pages, embeddable badges. Cached 60s.
- 48 Shipped ETA · Shipped May 2026
Accounts + sign-in
Email 6-digit code + GitHub OAuth. Magic-link-free, no passwords, no third-party identity provider in the request path. Sessions in HttpOnly Secure cookies.
- 42 Shipped ETA · Shipped May 2026
Discord / Slack / generic webhook alerts
Alerts where devs actually live. Add a webhook URL in profile prefs — we auto-detect Discord / Slack / generic JSON and POST a structured payload on every starred-service state change. HMAC-signed for the generic kind. Pro feature.
- 37 Shipped ETA · Shipped May 2026
Email alerts when a starred AI service degrades
Star the AI services you depend on. We email you the moment one flips state — outage, partial degradation, recovery. Configurable severity threshold so you only get paged for what matters. Pro feature.
- 32 Shipped ETA · Phases A-E shipped May 2026
AI news feed — releases, papers, tooling drops
Phases A-E SHIPPED: ingestion + /news + per-article + per-source pages, 12 curated sources, full-text mirroring with credit + link back, JSON-LD + sitemap entries. Engagement: likes, threaded comments with markdown + replies (depth-1) + 5-min edit window + soft-delete. Personalisation: subscriptions, dismissals, "For you" vs "All" tabs, heuristic ranking, /news?liked=1 filter. Discovery: tag pages + popular-tags cloud + free-text search. Distribution: /news.rss Atom feed + RSS auto-discovery in layout. Pro: instant news notifications when a subscribed source publishes (5-per-tick cap, 7d idempotency). Profile integration: subscribe list + recent likes + dismissal management. Digest emails carry Top 5 weekly/daily AI news. Future phase: R2 image caching (when traffic justifies it).
- 23 Shipped ETA · Shipped May 2026
Embeddable status badges + public API
Shields-style SVG badges for any tracked service — drop them into READMEs, docs, dashboards. Free public REST API at /api/v1/statuses with 60s cache. No auth required.
- 22 Shipped ETA · May 2026
Latency probe + uptime sparkline
Probe each provider on every cron tick, record response time, render a 24h sparkline on /service/[slug] + an aggregate ms tile on /status. Real signal we own, not delayed status-page updates.
- 14 Shipped ETA · May 2026
Daily / weekly status digest email
Opt-in summary of every incident across your starred AI services in the last 24h (daily) or 7d (weekly). TZ-aware send-time picker, multi-destination fan-out (up to 5 verified addresses), per-destination theme. Pro feature.
- 13 Shipped ETA · Shipped May 2026
Cron-driven snapshot refresh (no traffic dependency)
Status snapshots used to refresh only when traffic hit /status on cache miss. On a low-traffic site that meant alerts could lag. /api/internal/refresh-status is now hit by an external cron every 5 min — diff runs reliably, regardless of who is viewing the page.
- 0 Shipped ETA · Shipped May 2026
Anonymous per-service subscribe landing
SEO landing at /subscribe/{id} for searches like "subscribe to anthropic status". HMAC-signed confirm-by-email flow, no account required. Drops into the same alert pipeline starred services use, with a one-click unsubscribe token. Public, free.
- 0 Shipped ETA · Shipped May 2026
Admin analytics dashboard
In-app /admin/analytics tab with KV-backed metric tiles (users, Pro, online, signed-in last 7d, email-verified, GitHub-linked, services, cron health), cron + digest heartbeat detail, and one-click links out to Cloudflare Web Analytics, KV, Pages, GitHub Actions, Resend, and Google Search Console. Operator-only.
- 0 Shipped ETA · Shipped May 2026
Diff-only KV writes — stay under CF free-tier quota
The cron path used to write ~8,400 status-cache entries per day (8× the 1,000-write Cloudflare KV free-tier ceiling). rememberLastGood + setLastIndicators + the snapshot write now read existing entries first and skip when nothing material changed. Steady state dropped to ~600 writes/day with one extra read per probe.
- 0 Shipped ETA · Shipped May 2026
AI crawler discovery — robots.txt allow-list + llms.txt
Explicit User-agent blocks in robots.txt for GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended, CCBot, and friends. A curated /llms.txt summarising the site for answer engines (ChatGPT browsing, Claude web search, Perplexity, Gemini) so Prismix data gets cited verbatim rather than synthesised from stale copies.
- 0 Shipped ETA · Shipped May 2026
Per-service incident history page
Extended 30-day incident history at /service/{id}/incidents with summary tiles (active / total / major-critical / maintenance), permalinks to each incident, breadcrumb nav, and JSON-LD ItemList + BreadcrumbList for long-tail SEO ("anthropic outage history", "openai incidents may 2026").
- 0 Shipped ETA · Shipped May 2026
Public incident RSS / Atom feed
/incidents.rss — every incident across the 48 tracked services, newest first, with provider + impact tags. Subscribe in any RSS reader; pipe into a Discord webhook RSS bot; index into your own ops dashboard. Public, free, no auth.
- 0 Shipped ETA · Shipped May 2026
/incidents — cross-service timeline page
A first-class page listing every incident across every tracked AI service, with permalinked per-incident pages, JSON-LD Event schema, and an Atom feed companion. Same data the /status feed surfaces, but as a paginated, SEO-friendly view.
- 0 Shipped ETA · Shipped May 2026
Quiet hours — pause alerts during sleep / weekends
TZ-aware time-of-day window. When the user's local hour sits inside [start, end), transition alerts drop silently for that tick. Wraps midnight. v1 is silent-drop; v2 will queue + deliver a morning summary instead. Pro feature.
Have an idea we missed?
Submit it to the queue. Other users vote — the ones that gain traction get prioritised on the actual build schedule.
Sign in to submit + vote
Free account, 30 seconds. Email code or GitHub one-click. We bring you right back here.