Triggering an analysis
Five ways to start an analysis — dashboard, API, chat, support ticket, watchdog rule — and when to use each
There are five ways to trigger an analysis. They all produce the same result schema; the difference is where the input comes from.
The five triggers
| Trigger | Best for |
|---|---|
| Dashboard | Investigating something you noticed manually |
| API | Programmatic invocation from CI, scripts, or your own tools |
| Chat command | Triaging during an incident in Slack / Teams / Discord / Google Chat |
| Support ticket | Auto-triage of incoming tickets in Zendesk / Freshdesk / Intercom |
| Watchdog rule | Proactive auto-analysis when a monitoring rule fires |
Dashboard
Project → Analyses → New analysis.
| Field | Notes |
|---|---|
| Description | What's wrong, in your own words. Short is fine — "checkout errors" |
| Time window | When to look. Pick from the presets (last 15m / 30m / 1h / 6h / 24h) or set a custom window |
| Screenshot | Optional. Drag-drop or paste — see Screenshots |
Click Analyze. Result appears in 30–90 seconds.
API
POST /v1/analyses with a JSON body:
{
"description": "checkout errors",
"timeStart": "2026-04-25T14:00:00Z",
"timeEnd": "2026-04-25T14:30:00Z",
"projectId": "proj_..."
}Authenticate with an API key in the Authorization: Bearer <key>
header. Response includes the analysis ID; poll GET /v1/analyses/{id}
or set up a webhook to be notified on completion.
See the API Reference for the full schema.
Chat command
@SigSentry analyze <description> <time-expression> [--project=<slug>]
Example:
@SigSentry analyze "checkout errors" last 30m --project=checkoutThe diagnosis posts as a card response in the same channel within a minute. See Chat commands for the full grammar.
Support ticket
If you've connected a support desk, every incoming ticket triggers a triage step automatically. High-severity tickets can also auto-trigger a full analysis based on your configured threshold. The diagnosis is posted back into the ticket as an internal note.
You don't trigger these manually — the support desk integration handles them as tickets arrive.
Watchdog rule
Watchdog rules with action: auto_analyze trigger a full analysis
when their conditions are met. The lookback window of the rule is
used as the analysis time window, and the rule's name + match details
become the description.
You don't trigger these manually either — the rule fires on schedule.
Picking the right time window
The single highest-leverage choice you make when triggering an analysis. Tighter windows yield cleaner diagnoses. See Time windows for strategies.
What gets sent
Whichever trigger you use, three things go into the analysis:
- Your description (free text)
- The time window (start and end)
- The screenshot if attached (dashboard or API only)
Everything else — log evidence, code correlation, project context — SigSentry assembles automatically from your project's configuration.