Changelog

What's new in PulsAPI

New features, improvements, and fixes. We ship fast and keep you informed.

Ready to useShipped in v1.16.0

The PulsAPI MCP server is live

Point Claude Code, Claude Desktop, Cursor, VS Code, or any MCP client at one URL and it gains five read-only tools for live status across 2463+ vendors. There is no API key to paste: the client registers itself, you approve it once in the browser, and the config file you commit holds no secret.

claude mcp add --transport http --scope user pulsapi https://www.pulsapi.com/api/mcp

Developer hub MCP documentation Read the announcement Connect a client

Read-only tools over public vendor status. MCP access is on the Pro plan and above.

Bug Fixv1.17.0September 19, 2026

Measured Numbers Only, Corrected Uptime & a Rebuilt Free Plan

Every figure on the site is now something PulsAPI measured, or an em dash. Uptime stopped counting rows that were never availability signals, AWS's destroyed Middle East regions stopped reading as healthy, and the free plan is three vendors watched properly rather than a trial of the whole product.

Free is now 3 monitored services with 15 days of incident history, and every alert for those three with no monthly notification cap. A monitoring product that goes quiet because a quota ran out is worse than no monitoring product, so there is deliberately no notification limit
Every paid tier is unlimited monitors, permanently: the meter comes off at $29 and never comes back, rather than stepping you through 25, then 75, then 150 as your stack grows. Prices are unchanged at $29 and $99, and Enterprise now publishes a floor (from $499) instead of reading as unpriced
History is the second axis: 15 days free, 90 days on Pro, 13 months on Business. 13 months exactly, because that is what the retention job keeps, so no plan promises rows a nightly job has already deleted
Nothing is deleted when a plan changes. Monitors past the cap stay visible and stop alerting, and every one of them returns on upgrade; accounts from before this change keep their original 25-monitor allowance
The trial-ending email now names all three things that actually change at expiry, not just the monitor count: monitors, alert channels beyond email, and history dropping from 13 months to 15 days. It also says plainly that nothing is deleted
Plan data derives from one file. The pricing page, the landing section, pricing.md, llms.txt and both JSON-LD blocks now render from the same table, which had already drifted: the structured data an AI assistant reads was serving $59 and $149 long after the page said $29 and $99
Uptime no longer counts rows that were never availability signals. Some vendors publish one component per tradeable asset or per edge PoP and mark delisted ones permanently down; those readings are correct, but they dragged one venue's published uptime to 23.91%. 2,177 catalog components across five providers keep their real status and stop voting in the service mean, which puts that venue at 86.93%
The most serious fix in this release: an unresolved AWS event that had gone quiet for 24 hours was being dropped, which reported the destroyed me-south-1 and me-central-1 regions as fully operational, flipping 271 components green on infrastructure AWS says it cannot restore. That horizon now governs the history feed only, an event AWS is still serving as current ends by resolution rather than by silence, and the 271 false readings were purged
A bulk correction that rewrote 493 long-impaired components to UNKNOWN was applied and then rolled back in full. Reading each event's update log rather than its start date showed Bahrain and UAE really are gone, with AWS posting updates as recently as September 15, so AWS's 95.37% is a real number. Every other provider checked matched its live status page
Incident timestamps: Salesforce incidents took their start from when Salesforce filed the report rather than when impact began, which made 56% of them end before they started, and Adobe recorded 1970-01-01 whenever a start was unknown. Both now read the real impact window, and 360 epoch-dated incidents were removed
Incident pages no longer print negative durations. Where a vendor's own feed reports a resolution before the start, the page shows both timestamps and omits the duration instead of rendering -16h 23m
Service pages report real region counts. AWS was publishing 6,103 regions by reusing its component count; the API now returns distinct regions (36 for AWS), and services probed as a single endpoint are no longer asked the region question at all
The landing page does not invent numbers any more. When the API cannot be reached, the operational count and every uptime figure render as an em dash instead of falling back to constants: a hardcoded 2,412 of 2,463 operational, a 99.97% catalog average that was never the catalog, and eight green vendor rows carrying made-up percentages
The hero board now asks for the eight vendors it names, by slug, instead of hoping they appear in the first 60 rows of an alphabetical catalog of 2,463 (only AWS did), and its uptime column is labelled with the window it actually reports
The polling claim is the true one now. A flat every 60 seconds was wrong for roughly 83% of the catalog. PulsAPI re-checks a vendor every 60 seconds whenever it is not fully operational or has changed state in the last 30 minutes, and every 5 to 20 minutes while it is stable, so an incident is always on the fastest cadence. Corrected on the hero, /about, /developers and the FAQ that ships as structured data on every service page
Blog posts and /about carry the one real byline. Four invented authors with job titles and biographies had been bylining the blog and emitting schema.org/Person markup on every post
Fixed a UI stall where an animated swap nested inside an animated parent never completed: the pricing billing toggle went on showing annual prices after switching to monthly, and the landing Fusion Engine ignored every click

New Featurev1.16.0September 12, 2026

A Ready-to-Use MCP Server, OAuth 2.1 Sign-In & a Rebuilt Admin Console

Point any MCP client at one URL and your AI agent can read live vendor status. You approve it in a browser instead of pasting an API key, because PulsAPI now runs its own OAuth 2.1 authorization server.

Remote MCP server at /api/mcp, ready to use today: Streamable HTTP, JSON-RPC 2.0, protocol revision 2025-06-18, and stateless, so there is no session handshake to negotiate
Five read-only tools an agent can call: list_services, get_service_status, get_component_status, list_incidents, and get_uptime_report, covering every tracked vendor
Connect with a URL and nothing else. Claude Code, Claude Desktop, Cursor, and VS Code run the authorization flow themselves; stdio-only clients such as Codex reach it through the mcp-remote bridge
PulsAPI now runs a full OAuth 2.1 authorization server: authorization code with mandatory S256 PKCE, RFC 7591 dynamic client registration, RFC 8414 and RFC 9728 discovery, and RFC 8707 audience binding so a token minted for the MCP endpoint is worthless anywhere else
Consent screen at /oauth/authorize, and a connected-clients list in Settings, MCP server where any client can be disconnected without rotating a secret
Short, rotating credentials: 60-second authorization codes, one-hour access tokens, and 30-day refresh tokens that rotate on every use, with the whole chain revoked if a spent token is replayed
The anonymous MCP tier has been removed. Every tool now requires a token and the Pro plan, which is what makes per-account rate limits, revocation, and an audit trail possible; its address-hashing rate limiter moved to guard registration and token exchange instead
API keys remain the right answer for headless callers: a key with the mcp:invoke scope works over the same endpoint for CI, where no browser is available to approve a prompt
Admin console rebuilt around the catalog and the ingestion pipeline: six sections now load independently instead of blocking behind the slowest one, and the integrity audit now reports distinct problems instead of restating one problem 40,000 times
Free plan restructured to 5 monitors with Slack delivery moving to Pro; existing accounts above the new limit are grandfathered and keep what they have
Public service pages no longer fan out roughly 40 queries per request, which was starving the connection pool and slowing the whole site, not just that one endpoint
Observability cleanup: client disconnects and cancelled probe timeouts no longer file Sentry issues for events the application handled correctly
Deploy safety: the V149 history rollup no longer backfills on the deploy path, where the healthcheck window bounds startup and not the work itself

Improvementv1.15.0August 19, 2026

Tiered Crawling, Component Fleet View & Region Flags

Popular services now refresh several times faster without adding crawl load, service pages describe every tracked component instead of the first twenty, and regions carry their country flag.

Tiered probe scheduling: high-demand services refresh roughly five times more often than the old flat sweep, while the dormant long tail backs off, same total request budget
Anything mid-incident or that just changed status is escalated to the fast tier automatically, so recoveries surface quickly and flapping stays visible
Conditional GET (ETag / Last-Modified) caching against vendor status pages, unchanged payloads no longer cost a full fetch and parse
Service pages now report the full component fleet: total tracked count and a status histogram, with the visible list ordered worst-status-first instead of alphabetically
Country flags on region rows and component regions, served from our own origin rather than a third-party CDN
CrowdStrike Falcon monitoring added via authenticated-surface probing across nine Falcon clouds (CrowdStrike publishes no public status page), catalog now 2,463 services
Sitemap correctness: per-URL lastmod derived from real status activity, and full dynamic generation so no page ships a frozen timestamp
Uptime accuracy fix: per-component uptime now resolves the latest reading per component instead of mixing in stale rows
Anthropic and Claude merged into a single service entry, no more split incident history across two pages

New Featurev1.14.0August 15, 2026

Teams, SAML SSO, Roles & New Pricing

Workspaces get real access control: team-owned dashboard groups, Owner/Admin/Member roles, and SAML single sign-on with optional enforcement, alongside a rebuilt pricing model and alert rules that can target a single component.

Dashboard groups are now owned by the team rather than shared copy-by-copy, everyone on the team sees the same board, and access follows membership
Role-based access control across teams, groups, boards, and SSO settings with Owner, Admin, and Member permissions enforced server-side
SAML 2.0 single sign-on: configure your IdP under Teams → SSO, with metadata exchange, assertion replay protection, and optional SSO enforcement for the whole team
Seat meter and team switcher in the app shell, see remaining seats before inviting and move between workspaces without a reload
Rebuilt invite flow with clearer join, expiry, and already-a-member states
New pricing: Free now includes 25 monitors, and monitors are unlimited from Pro up, no more upgrading a tier just because your stack grew
Plan limits now return structured errors, so hitting one opens a specific upgrade dialog naming the limit instead of a generic failure
Alert rules can be scoped to a specific component or dashboard group, not just a whole service
Group members now receive board alerts, previously only direct subscribers were notified for group-added services
Namecheap status monitoring added

New Featurev1.13.0August 9, 2026

2,400+ Services, Demand-Ranked Directory & Privacy Controls

The catalog grows from 292 to over 2,400 monitored providers, the directory ranks by real demand instead of alphabet, and analytics now runs behind explicit consent.

Service catalog expanded from 292 to 2,400+ monitored providers across cloud, SaaS, data, developer tooling, security, and consumer platforms
Directory ranked by demand so the providers people actually check surface first, with category pages and a faster server-side status rollup
First-party logos for the expanded catalog, served from our own origin, with a strict no-guessing rule so a wrong logo is never shown
Onboarding catalog picker: search the full catalog and build your board during connect-stack instead of adding services one at a time afterwards
Analytics behind Consent Mode v2, nothing tracks until you opt in, with granular controls under Cookies and Privacy Choices
Settings and category UI refinements across the app shell, plus a redesigned directory table
Reliability: fixed a cache serialization bug that could 500 every status surface, and corrected per-IP rate limiting that counted all server-rendered traffic as one client
Hetzner, CometVPN, Mailgun, and CloudFront monitoring fixes; AWS and Google component mapping corrections

Improvementv1.12.0July 23, 2026

Catalog Growth, Homepage Copy & Docs Refresh

The service directory reaches 292+ providers, homepage messaging is clearer and more direct, Fusion Engine samples get syntax highlighting, and product docs get a full copy pass.

Service catalog expanded to 292+ monitored cloud and SaaS providers with refreshed registry descriptions
Homepage copy refresh: sharper hero, FAQ, stack health, and pricing messaging
Fusion Engine: lightweight JSON and XML syntax highlighting on vendor payload samples
Product documentation copy pass across getting started, dashboard, alerts, integrations, SLA, teams, and support
AI agent skill and llms.txt aligned to the current catalog size and PulsAPI naming

New Featurev1.11.0May 17, 2026

Auto-Discovery, Resilience, Postmortems & Executive Reporting

Wire Stack Health from real infrastructure and repos, score resilience over time, close the loop with AI-assisted postmortems, and give leadership a concise readout, plus a public status wall that shows every vendor at a glance.

Capability auto-discovery: connect AWS and GitHub, run scans, and review suggested capabilities and dependencies before they land in Stack Health
Resilience scoring per capability with history, plus aggregate stack health and on-demand recomputation when your model changes
Incident verification so teams can record what was checked during an outage, improving downstream resilience and reporting quality
Postmortems workspace: generate drafts from incident context, iterate, publish, and export to Markdown or PDF
Executive reporting (Pro): distribution, summaries, and rollups under Settings → Executive Reports for stakeholder-ready snapshots
Public Cloud Status Wall (/status-wall): live grid of vendor statuses in one place, with an embed mode for displays and iframes
Homepage copy and layout refresh across hero, features, and trust sections
Performance indexes on hot paths so dashboards stay responsive as monitored services and history grow

Improvementv1.10.0May 7, 2026

AWS Flagship Products: Dedicated Service Pages

EC2, S3, Lambda, RDS, DynamoDB, Route 53, EKS, ECS, IAM, CloudWatch, SQS, and SNS each get their own PulsAPI service page, their own incident timeline, SLA metrics, components, and subscription.

EC2, S3, Lambda, RDS, DynamoDB, Route 53, EKS, ECS, IAM, CloudWatch, SQS, and SNS now have dedicated PulsAPI service pages (slugs aws-ec2, aws-s3, aws-lambda, …)
Per-product incident timeline, SLA report, component breakdown, and subscription, no more navigating one giant AWS page
Other AWS products (API Gateway, Cognito, Glue, …) continue to aggregate under the generic AWS service
UI fix: top-level service status now correctly reflects the worst component state across all regions
CloudFront component mapping fixed; AWS region-suffixed component keys grouped properly

New Featurev1.9.0May 4, 2026

Expanded Service Coverage & Reliability

42 new services in the directory, Hugging Face and Mercury added, AI search visibility improvements, and a wave of crawler and account reliability fixes.

42 new services added to the directory, broader coverage across cloud, data, and developer tooling
Hugging Face status monitoring added; Mercury direct health check reactivated
Improved AI search visibility (GEO/AEO): quotable answer blocks and FAQ schema across content pages
Blog index polish and CSV export fix for incident history downloads
Crawler reliability: status normalisation fixes, Google Search Console URL and sitemap corrections
Account and password flow refinements; admin audit log and metrics dashboard polish
Cache headers tuned for service detail pages for faster repeat loads

New Featurev1.8.0April 20, 2026

Stack Impact Intelligence & Expanded Service Coverage

Map your actual stack to live vendor status and see what's at risk in business terms, plus 21 new services added to the directory.

Stack Impact Intelligence: model business capabilities, their dependencies, and environment scope (prod/staging/dev) so incidents translate into concrete business impact
Capability dependency graph with criticality levels and assessed severity scoring as upstream vendors change status
Capability impact history to review which parts of your stack were affected, when, and for how long
Component search across all providers to quickly wire dependencies to the right upstream component or region
Canonical regions API: unified region identity across AWS, Azure, GCP, and Atlassian for cleaner cross-cloud reporting
21 new services added to the directory, bringing broader coverage across cloud, data, and developer tooling
SEO enhancements across service and marketing pages, logo and description fixes, and crawler reliability improvements

New Featurev1.7.0April 11, 2026

Integrations Hub, Embeds, Firehose & MCP Server

A complete integration layer: 26+ outbound connections, live status embeds, real-time firehose streaming, and an AI/MCP server for natural language vendor queries, plus a full features comparison page.

Integrations Hub: 26+ integrations across chat (Slack, Discord, Teams, Google Chat, Webex), incident management (PagerDuty, OpsGenie, Better Uptime, FireHydrant), helpdesk (Zendesk, Freshdesk, Freshservice, Incident IQ), cloud (AWS, Azure, GCP, Microsoft 365), and notifications (email, SMS)
Dedicated marketing pages for every integration with setup guides, key features, use cases, and step-by-step HowTo schema
Status Embeds: four embed types, iFrame, inline widget, modal, and TV dashboard for NOC screens
Firehose: real-time webhook streaming for all status events, ideal for custom pipelines and data warehouses
MCP Server: native Model Context Protocol support so AI tools like Claude can query live vendor status in natural language
Settings integrations tab redesigned, connect, configure, view sync status, and review delivery history per integration
Features page with a full competitor comparison table: PulsAPI vs StatusGator, Instatus, Better Stack, and Atlassian Statuspage across 9 dimensions
Sitemap updated with /features and all 26 /integrations/[slug] pages; FAQPage and WebPage JSON-LD added to features page for rich results eligibility

New Featurev1.6.0March 28, 2026

Alert Rules, Escalation & Quieter Incidents

Define when and how you get notified, step up urgency over time, and pause alerts during planned maintenance, with history and stats to review what fired.

Alert rules engine: thresholds, status transitions, consecutive failures, downtime budget, SLA breach, incident, and recovery conditions
Escalation policies with ordered steps, delays, and repeat options so the right channel gets pinged if something stays broken
Maintenance windows to suppress alert noise while vendors publish planned work
Alert event history plus daily statistics for a clear picture of what triggered and when
Flap detection to dampen alerts when status flickers
Broader status page coverage and fetchers (e.g. Firebase, Hetzner, Webex) alongside ongoing crawler reliability work

New Featurev1.5.0March 10, 2026

Blog & Content Hub

We launched the PulsAPI blog, engineering insights, status monitoring best practices, and product deep-dives.

New /blog section with articles on cloud reliability and operational intelligence
Dedicated changelog page (you're looking at it)
Redesigned professional footer with all navigation links
Consistent marketing header across all public pages

New Featurev1.4.0February 18, 2026

Team Collaboration

Invite team members, assign roles, and collaborate on shared dashboards.

Team invitations via email with tokenized join links
Role-based access: Owner, Admin, and Member permissions
Shared dashboard groups visible to all team members
Up to 5 team members on Pro, unlimited on Enterprise

New Featurev1.3.0January 25, 2026

Custom Dashboard Groups

Organize your monitored services into logical groups, by project, team, or infrastructure layer.

Create unlimited custom dashboard groups on Pro plans
Drag-and-drop service ordering within groups
Group-level status aggregation (worst-status-wins)
Quick-switch between groups via the dashboard sidebar

New Featurev1.2.0December 12, 2025

Smart Alert Routing

Route incidents to the right people, on the right channel, at the right severity level.

Slack integration: receive alerts in any channel via incoming webhooks
Discord integration: webhook-based notifications with rich embeds
Per-service, per-severity routing rules (e.g., AWS critical → Slack #infra, Stripe degraded → email)
Alert deduplication to prevent notification storms during cascading incidents

New Featurev1.1.0November 5, 2025

SLA Metrics & Performance Tracking

Hard numbers to hold vendors accountable, 90-day uptime, MTTR, and incident frequency.

90-day rolling uptime percentage for every monitored service
Mean Time to Recovery (MTTR) calculations per provider
Incident count and severity breakdown over time
Performance dashboard with day-by-day uptime timeline
Exportable SLA reports for vendor reviews

Bug Fixv1.0.1October 20, 2025

Reliability Improvements

Bug fixes and stability improvements to the crawler and notification pipeline.

Fixed crawler not updating status on transient network failures
Resolved email notification delays during high-incident periods
Improved service icon resolution for 40+ providers
Fixed pagination edge case in the service directory

Launchv1.0.0October 1, 2025

PulsAPI Launch

The first public release of PulsAPI, unified cloud status monitoring for engineering teams.

Real-time status monitoring for 2463+ cloud services across 8 categories
Unified dashboard with live status tiles and last-updated timestamps
Service directory with search, filtering, and uptime display
Incident timeline with chronological status updates
Component-level granularity (e.g., AWS S3 us-east-1, Lambda eu-west-2)
Email notifications for status changes
GitHub OAuth and email/password authentication
Free plan: up to 10 services, 30-day history
Pro plan: unlimited services, 90-day SLA, advanced alerts