radish is a browser-side incident analysis tool. Paste any combination of log lines, stack traces, Slack messages, or error output — radish extracts a timeline, classifies the failure type, and generates a structured postmortem with root cause, 5 whys, contributing factors, and action items.
Supported failure modes include OOM crashes, connection pool exhaustion, post-deploy regressions, authentication failures, timeout cascades, crash loops, and config drift. The heuristic engine runs entirely in your browser — no data is uploaded and the tool works offline once the page has loaded.
Output can be shared as a permanent link (state encoded in the URL hash) or copied as a Markdown postmortem. Use it during an incident for fast framing, or after to write the blameless postmortem before memory fades.
Is my incident data uploaded? No. All analysis runs in your browser using JavaScript pattern matching. Nothing leaves your device.
What input does radish accept? Any free-form text — logs, stack traces, Slack threads, monitoring alerts, or a stream-of-consciousness incident summary. Paste whatever you have.
How accurate is the analysis? radish uses heuristic pattern matching, not an LLM. It is fast and private, but works best when the input contains recognizable signals (timestamps, error codes, error keywords). Complex or ambiguous incidents may need manual review.
Can I share the results? Yes — click "copy link" to get a URL that encodes the full analysis. Anyone with the link can view the postmortem without signing up.
Is this free? Yes, completely free. No account required.