---
title: Replay-Driven Triage
description: Use Luotsi when the main question is how to triage Android automation failures from artifacts after the run has already ended.
---

If you searched for replay-driven triage, the important distinction is whether the artifact root is good enough to answer the next question without reconnecting to the device. Luotsi is built around that assumption.

## Why the artifact root matters

Luotsi writes replay-friendly material so the session can be examined after the device is gone:

- `session-timeline.jsonl`
- `session-replay.json`
- screenshots and hierarchy captures
- logcat and metadata
- report and failure bundle outputs

That lets triage stay host-side and asynchronous instead of forcing another live session just to see what happened.

## Recommended commands

```bash
luotsi replay packet --artifacts ./artifacts/<run>
luotsi replay packet --artifacts ./artifacts/<run> --check
luotsi replay open --artifacts ./artifacts/<run> --dry-run --write-json --write-markdown
luotsi replay summarize --artifacts ./artifacts/<run>
luotsi replay search --artifacts ./artifacts/<run> --contains "error"
luotsi replay capsule --artifacts ./artifacts/<run>
```

Start with `replay packet` when CI or an agent needs the durable packet directly: it writes `run-summary.json` and `run-summary.md`, returns the same `luotsi-run-summary.v1` object, includes the At a Glance summary, failure snapshot, primary failure, and recommended next action, and never launches a browser. Validate shared packets with `replay packet --check`. Use `replay open --dry-run --write-json --write-markdown` when a human also wants the replay front-door response and browser entrypoints.

## Strong fit when

- The device session is expensive or impossible to repeat immediately.
- Engineers or agents need to search, summarize, and inspect artifacts after the run.
- You want failure evidence that can move between local debugging, code review, and CI.

## Weaker fit when

- The workflow depends entirely on live interaction and leaves no meaningful artifacts behind.
- Your team only needs a pass or fail bit from CI and never revisits failed runs.

## Open these next

- [Replay and Artifacts](../../core-workflows/replay-and-artifacts/)
- [Replay Graph And Clusters](../../reference/replay-graph-and-clusters/)
- [Output Envelopes](../../reference/output-envelopes/)
- [AI Agent Workflows](../../core-workflows/ai-agent-workflows/)