Every part of errorgap writes into the same place. Start anywhere — errors, logs, uptime, performance — and the signals meet on one incident with a likely cause.
Six heuristics promote error spikes, failed checks, and latency anomalies into one incident with a ranked likely cause.
/product/incident-intelligence →Grouping you can readFingerprints group same-cause errors, with a readable rationale, merge/split, and source context per frame.
/product/backtrace →Stream · list · aggregateThe errorgap-agent streams production logs live — stream, list, and aggregate views, duplicates grouped.
/product/logs →HTTP · TCP · ping, 10sHTTP(S), TCP, and ping checks every 10 seconds, with a 90-day availability timeline per monitor.
/product/uptime →Per-transaction timingPer-transaction timing broken into DB, view, and external components. Latency spikes feed the incident engine.
/product/apm →Typed event timelinesTyped event timelines rather than video, attached to error groups and incident evidence.
/product/session-replay →Production context for AIAn MCP server with eight tools, six-category exposure controls, and regex redaction before anything leaves.
/product/agent-access →GitHub · Slack · Jira · …GitHub, GitLab, Jira, Slack, and webhooks — with an issue composer and PR → release → verification tracking.
/product/integrations →Correlation is the reason the hub is more than a link list. Each capability sends a different kind of signal; the incident engine reads them together on a 60-second cycle.
What an incident carries by the time you open it. Nothing here was assembled by hand.