Pulsimo is in beta and changes often. These notes come from each feature page and list what was fixed or added, newest first.

17 Oct 2026#

API metrics and Event trail#

The Event trail used to show the events of every organization on the server. It is now limited to your organization: each event is tied to its organization through what it is about (incident, endpoint, source…) or who did it, and older events were backfilled (migration 038).

Read API metrics and Event trail.

14 Oct 2026#

Projects and overview#

The old pages had several wrong or invented figures:

Total projects (1) and the SLA list (2) disagreed.
Projects Analytics showed 100% uptime next to 200 incidents, and Issues 0 for every project.
Avg detection time 2–5 min was fixed text.
The trend arrows showed the uptime level, not a change over time.
The incident figures came from the latest 1000 incidents, whatever the period.

All figures now come from the checks and incidents of the selected period, and the trend compares with the period before. A reliability target per project is new (migration 036). The Services overview is now a live status board, replacing the old performance table.

Read Projects and overview.

Channels and delivery#

The channel form was cut off at the bottom. It now scrolls, with the buttons always visible.
Send test used a separate code path in the gateway. That path ignored the email settings in .env and didn't show why it failed. It now goes through the notification service, the same as real alerts, and shows the platform's answer.
Paused channels used to disappear from the page. They now stay listed, with Pause and Resume.
Deliveries now record their endpoint. The delivery log, the per-channel figures and the Failed figure are new.
Reminders were timed per channel, so two endpoints down at once delayed each other's reminders. They are now timed per endpoint.
Webhook URLs and header values are masked in the page and the API.
Viewers can no longer change channels, and only administrators can delete them.
Slack, Discord and Teams messages are colour-coded. Webhooks now use the method you chose (it was always POST). Email supports several recipients and port 465.
The SMTP fields in the email form were removed. They were never used: email always went through .env.

Read Channels and delivery.

Settings and API keys#

Session & Authentication didn't work.
- The number fields reset while you typed, so you couldn't enter 30 or 120.
- The lock duration could not be set.
- "Never" could not be chosen for password expiry once a value was saved.
- Errors only said Request failed with status code 400.
- The "session timeout" signed everyone out a fixed time after signing in, even in the middle of work.
It is now an inactivity timeout with a warning. Every field accepts typing, and errors say what to fix.
Settings that did nothing were removed:
- Security: two-factor, IP allow-list, SSO, "encrypt sensitive data", "automatic redaction", the audit logging switch, and the compliance badges;
- Advanced: "experimental features" (including an AI switch) and a "reset everything" button that did nothing.
API keys work. Before, keys were created but every request using one was refused. Keys now have access levels, expiry, last-used and creator, and only administrators manage them.
The Integrations tab used to show placeholder cards (Datadog, New Relic, Zapier…) that couldn't connect to anything. It now shows real connections.

Read Settings and API keys.

12 Oct 2026#

Service analytics#

Organization scoping. These APIs used to return every organization's endpoints; they now return only yours.
Apdex is exact. It was summed from overlapping one-hour windows written every minute, which counted each check about 60 times (703,540 "slow requests" against about 32,500 real checks). It is now computed from the checks themselves, with failed checks counted as frustrated. The checker's hourly Apdex table now uses one row per clock hour.
Anomalies no longer fail. Traffic and 5xx anomalies from metrics broke the anomalies API (HTTP 500), and the health-check anomaly detector had never run. Both are fixed.
Network timing is labelled honestly. The five phases describe a first request; the check's own response time is shown next to them.

Read Service analytics.

Users and access#

Access is enforced at once. A deactivated person kept working until their session expired, and could keep renewing it for up to 7 days. A role change only applied after signing in again. Both now apply on the next request.
Real numbers instead of placeholders. The Compliance 95% score (fixed values), the permission count (always 0), failed sign-ins (always 0) and the password age (taken from the account's creation date) are replaced by real data. Every request used to overwrite last sign-in, so it really meant last seen; the two are now separate .
User details work. The details request read columns that don't exist and always failed.
Safety rules (above) are new, as are unlock, sign out everywhere, the password policy on every password, and password history.
The Event Trail now records each user event under its organization. Before, user and other gateway events were saved without one, so they didn't appear in the organization's Event Trail.

Read Users and access.