Incident Timeline Reconstruction

Rebuilds a precise, source-attributed incident timeline from chat logs, alerts, and deploy records, separating what happened from when it was noticed and flagging gaps.

License: CC0-1.0 · By waxmark · v1 · 2026-11-02

Share on X

Add a trust badge to your README

[![Waxmark](https://waxmark.app/api/badge/incident-timeline-reconstruction)](https://waxmark.app/items/incident-timeline-reconstruction/scan)
Report
Reconstruct the timeline of the incident from the raw sources below. The timeline will anchor a blameless postmortem, so precision about time and source matters more than narrative; every entry must be traceable to a line in one of the sources.

<sources>
<document index="1">
<source>chat_log</source>
<document_content>
{{CHAT_LOG}}
</document_content>
</document>
<document index="2">
<source>alerts</source>
<document_content>
{{ALERTS}}
</document_content>
</document>
<document index="3">
<source>deploy_log</source>
<document_content>
{{DEPLOY_LOG}}
</document_content>
</document>
<document index="4">
<source>metrics_notes</source>
<document_content>
{{METRICS_NOTES}}
</document_content>
</document>
</sources>

Output all times in {{TIMEZONE}} (use UTC if empty), in `YYYY-MM-DD HH:MM:SS` format. Sources may use different zones and formats; convert, and state the conversion you applied for each source.

Procedure:

1. Before writing the timeline, extract every timestamped event from each source into a scratch list, quoting the original line and noting its source and index. Work through one source at a time to avoid blending them.
2. Normalize timestamps. Where a source lacks a date or zone, infer it from surrounding entries and mark the entry `[time inferred]`.
3. Classify each event as one of: `change` (deploy, config, flag, infra action), `symptom` (alert, metric shift, user report), `detection` (a human noticing), `action` (a mitigation attempt), `communication` (status update, escalation, page), `resolution`.
4. Merge into a single ordered timeline. When two sources describe the same event, keep one entry and cite both. When sources conflict on time, keep both times, mark `[conflict]`, and note the difference.
5. Derive the key intervals: time from first `change` to first `symptom`, first `symptom` to first `detection`, first `detection` to first `action`, first `action` to `resolution`. If any endpoint is uncertain, give a range.
6. List gaps: periods longer than 10 minutes with no entries during the active incident, decisions in the chat with no visible trigger, and actions referenced but never recorded as done. These are the questions for the postmortem meeting, not conclusions.

Describe actions neutrally: "deploy of build 4821 to prod started" rather than "engineer pushed a broken build". Attribute to roles or handles exactly as they appear in the sources; do not infer intent or blame.

## Output format

**Source normalization** - table: source, original time format and zone, conversion applied, number of events extracted.

**Timeline** - table with columns: time ({{TIMEZONE}}), type, event (one sentence, neutral), source and quoted fragment, flags (`[time inferred]`, `[conflict]`, or blank). Chronological.

**Key intervals** - table: interval name, start event, end event, duration (or range), confidence (`high` / `medium` / `low`).

**Gaps and open questions** - numbered list; each names the time window or event and the specific question to ask in the postmortem.

**Candidate contributing factors** - only factors directly supported by timeline entries, each citing the entry times. Label this section as hypotheses for discussion, not findings.
Incident Timeline Reconstruction · Waxmark