Changelog — OperatorMesh

v2.3.0 Major May 17, 2026

Blast Radius Predictor + Post-Mortem + Handoff — All Live

Three new world-first AI modes ship today. OperatorMesh now covers the full incident lifecycle — from pre-deploy scanning to live triage to post-incident documentation and shift handoff.

Why it matters: Engineers no longer need separate tools for each incident phase. OperatorMesh handles prediction, triage, documentation, and handoff in one place — all in seconds.

NEW💥 Blast Radius Predictor — if this service stays down, what else breaks? Cascade timeline at T+5/15/30 min, dependency chain, fix priority, and contain action. World first.

NEW📄 Post-Mortem Auto-Draft — instant blameless post-mortem from incident context. Timeline, contributing factors, action items with owners and due dates. World first.

NEW🔄 On-Call Handoff Briefing — 60-second shift briefing for incoming engineers. Status, open items, watch list, escalation contact. World first.

IMPDashboard mode tabs now include all 5 modes: Triage, Pre-Mortem, Blast Radius, Post-Mortem, Handoff

IMPMode-aware validation messages and button labels across all modes

IMPBlast Radius history cards show cascade timeline, dependency chain, fix priority inline

FIXPre-mortem validation no longer shows "incident" error message when in deploy scan mode

FIXButton label correctly resets to mode-specific text after analysis completes

⚠ No breaking changes. All existing triage history renders correctly.

v2.2.0 Major May 15, 2026

Rejected Hypotheses + Missing Signals + Bug Fixes

Explainability update — every triage now shows its reasoning, not just its conclusion. Plus webhook deduplication and dashboard reliability fixes.

Why it matters: "Explainable AI" means engineers trust the output and know exactly what evidence to gather next.

NEWRejected Hypotheses — every triage shows what alternative causes were considered and eliminated. World first in accessible incident triage tooling.

NEWMissing Signals — each triage identifies exactly what logs or metrics would raise diagnosis confidence. Turns "I don't know" into "here's what I need." World first.

NEWROI Calculator — interactive calculator at /roi.html. Enter incidents/month, triage time, engineer cost — see exact savings and ROI multiple.

NEWAccuracy Benchmark — 87% diagnosis accuracy across 50 incidents with failure cases and confidence calibration table.

FIXWebhook Deduplication — same alert firing 5× in 30 min now generates 1 Slack message. SHA-256 fingerprinting suppresses duplicates.

FIXResolution handling — PagerDuty/Grafana resolved events post a clean ✅ message to Slack and clear the dedup fingerprint.

FIXDashboard loading — parallel Supabase fetches, 5-second timeouts, proper loading state. No more infinite spinner.

FIXSign out — localStorage cleared on signout. Redirect works across all browsers.

IMPSupabase future-proofing — explicit GRANT statements ahead of October 2026 enforcement deadline.

⚠ No breaking changes.

v2.1.0 Major May 8, 2026

Pre-Mortem Scanner + Dual Confidence Scoring

Two world-first features that no incident triage tool has shipped accessibly before. Predict failures before they happen, and separate how confident we are in the diagnosis from how confident we are the fix will work.

Why it matters: Shift-left incident prevention stops outages before they start. Dual confidence scores give engineers honest signal about when to escalate vs act immediately.

NEWPre-Mortem Deploy Scanner — paste a git diff or describe a change, get predicted failure modes, at-risk services, post-deploy monitors, and a deploy safety score (0–100)

NEWDual Confidence Scoring — Diagnosis Confidence (how certain is the root cause?) and Remediation Confidence (how certain will the fix work?) shown separately

NEWEscalation warnings — auto-flag when remediation confidence drops below 60% with a plain-language warning in both the UI and Slack

NEWDashboard filter tabs — separate view for Triage history vs Pre-Mortem scan history

NEWPre-Mortem Scans stat card — track total deploys scanned on your dashboard

IMPSlack webhook messages now show Diagnosis and Remediation confidence separately with escalation alert when needed

IMPDashboard history cards expanded to show dual confidence bars with color-coded scores

⚠ No breaking changes. Existing triage history renders correctly using legacy single confidence score.

v2.0.0 Major May 3, 2026

Full platform: accounts, dashboard, integrations

OperatorMesh is now a complete workflow — from alert to resolution tracking — without switching tools.

Why it matters: Faster response, centralized visibility, and zero manual triage steps.

NEWUser accounts with email and magic link login

NEWDashboard with triage history and usage tracking

NEWAutomated Datadog and PagerDuty ingestion

NEWSlack delivery with root cause and actions

IMPFaster and more accurate analysis

SECAPI credentials secured server-side

FIXBilling reflects correct yearly pricing

⚠ No breaking changes.

v1.5.0 Update May 2, 2026

Automatic Slack delivery

Triage results now arrive in Slack instantly after an alert fires.

Why it matters: Eliminates manual debugging steps and reduces response time.

NEWSlack auto-delivery of triage results

NEWWebhook ingestion from monitoring tools

SECImproved credential security

IMPReduced pipeline latency

⚠ No action required.

v1.3.0 Update Apr 26, 2026

Analytics and notifications

Added privacy-first analytics and changelog notifications.

Why it matters: Understand usage trends without compromising user privacy.

NEWAnonymous analytics

NEWChangelog email notifications

FIXImproved footer usability

v1.1.0 Update Apr 23, 2026

Live triage experience

Users can now paste real logs and receive instant analysis.

Why it matters: Enables real-world usage without setup.

NEWInteractive triage widget

NEWSample incidents for testing

FIXInput validation improvements

v1.0.0 Launch Apr 19, 2026

OperatorMesh is live

First public release — focused on one thing: turning production incidents from 40-minute sessions into 3-second answers.

NEWoperatormesh.com — live and publicly accessible

NEWZero data retention by design

NEWNo agents, no installs, no infrastructure access

NEWWorks with Datadog, PagerDuty, Grafana, Sentry