ThoughtSpot Webhook MCP Server

ThoughtSpot Webhook

Deliveries land when a Liveboard schedule fires upstream. Each run is a LIVEBOARD_SCHEDULE event — often a JSON metadata envelope plus an export attachment (PDF, CSV, or XLSX). Tabular columns and KPI names vary by Liveboard, schedule, personalized view, and which visualizations were included; inspect each delivery rather than assuming a fixed schema.

First moves

1. Latest report — default for "what did the scheduled Liveboard export say".
2. Report freshness — check source.captured_at and source.filename before quoting figures.
3. Recent reports — compare runs, spot duplicate filenames, or find a specific schedule date.
4. One report by id — after listing, when the user names a specific run.

Response shape

• source — capture time, filename, format, connection purpose.
• export.rows — parsed CSV preview when tabular; may truncate with rows_truncated.
• export.text — full body when rows truncate or the file is not tabular.
• export.json — parsed JSON when the stored part was JSON metadata.

When a run arrives as multipart, storage may hold separate records for the metadata JSON (payload) and the attachment file — correlate by nearby captured_at timestamps. Prefer the attachment for metrics; use the JSON for schedule name, Liveboard title, runId, fileFormat, recipients, optional viewInfo, and optional aiHighlights.

Operational order

1. Start with the latest report — do not guess a report id.
2. If captured_at is older than expected, say so — upstream sends on the schedule cadence.
3. List recent reports only for history or explicit compare-by-date asks.
4. Read ## This connection in the skill supplement for Liveboard-specific column notes.

Gotchas

• Read-only — cannot trigger a new schedule run from here.
• Variable exports — CSV/XLSX columns reflect whichever Liveboard visualizations were scheduled; PDF exports may lack tabular preview — use export.text or note binary content.
• Multipart split — one upstream POST can produce two stored items (metadata JSON + file); pair them by capture time.
• Metadata-only runs — when upstream delivers files to cloud storage instead of the webhook, storage may hold JSON with file references rather than attachment bytes; tabular data may be absent.
• Personalized views — metadata may include viewInfo when the schedule targets a Liveboard view, not the full board.
• Deduping retries — compare msgUniqueId or runId in JSON metadata when upstream retries the same schedule fire.