name: thoughtspot-assessment
description: Take inventory of a ThoughtSpot instance and produce a migration-readiness readout — models/worksheets, Liveboards, Answers, tables, connections, per-object usage (views + distinct users from TS: BI Server), per-Liveboard chart-type mix and complexity, ownership concentration, Embrace-vs-Falcon data-source patterns, and a value/cost-ranked migration shortlist. Use to scope a ThoughtSpot→Sigma migration or audit BI sprawl. Read-only.
user-invocable: true
ThoughtSpot migration assessment
Read-only pre-scoping for a ThoughtSpot → Sigma migration. Complements
thoughtspot-to-sigma (which does the actual conversion). Designed for a real,
populated production instance — usage data is a first-class signal, not an
optional extra.
Auth
Same as thoughtspot-to-sigma: TS_HOST + TS_TOKEN (SSO session token or
Trusted-Auth service token). For the usage axis the identity needs admin scope so
it can read the TS: BI Server system worksheet. No Sigma credentials needed —
this only reads ThoughtSpot.
Run
scripts/scan.py — inventories the instance via metadata/search
(LOGICAL_TABLE / LIVEBOARD / ANSWER / CONNECTION), pulls per-object usage from
TS: BI Server, exports each Liveboard's TML, and per Liveboard scores migration
complexity: viz count + distinct chart kinds + models touched + TML formula and
filter weight. Writes ~/thoughtspot-migration/assessment.json.
Then render the customer-facing report:
ruby scripts/render-readout-html.rb --out ~/thoughtspot-migration
→ ~/thoughtspot-migration/readout.html (Sigma-branded, share-friendly).
scripts/render-readout-html.rb is the canonical renderer (the old
render_html.py is superseded — keep using the Ruby one for parity with the
tableau / powerbi / qlik / quicksight assessment skills).
What the readout covers
- 01 Environment overview — KPI tiles: Liveboards, Answers, models/worksheets,
tables, connections.
- 02 Liveboard priority & usage — usage-ranked with bars from
TS: BI Server;
cold / zero-view detection for retirement candidates.
- 03 Ownership & concentration — Liveboards by author; top-author concentration.
- 04 Data-source patterns — Embrace (live warehouse) vs Falcon (in-memory),
file-uploaded tables flagged for warehouse landing.
- 05 User activity — per-user action volume (only when
TS: BI Server usage
is present).
- Migration shortlist + per-Liveboard complexity — value/cost ranking, tag
pills (migrate-first / easy-win / needs-review / retire), chart-type coverage,
and the set of models the Liveboards depend on (migrate these first).
Usage data
On a populated instance, TS: BI Server (ThoughtSpot's built-in activity log)
yields per-Liveboard views + distinct users and per-user activity (admin scope).
This drives the value/cost shortlist and cold-content detection. If the worksheet
is genuinely absent (a brand-new instance, or a non-admin identity), the scan
records a single note and falls back to effort-only ranking — but the design and
the readout assume the populated case.
Chart-type coverage
The thoughtspot-to-sigma pipeline maps KPI / COLUMN / BAR / LINE / PIE / TABLE
/ ADVANCED_COLUMN / stacked / area / LINE_COLUMN to Sigma equivalents. Flagged
for review (no direct 1:1 in the current pipeline): SCATTER, BUBBLE, GEO_AREA,
PIVOT_TABLE, WATERFALL, FUNNEL, TREEMAP, LINE_STACKED_COLUMN. Sigma supports most
of these natively — they just need element-builder mapping work.
Notes
- TML export embeds raw control chars in JSON →
json.loads(..., strict=False).
- ThoughtSpot TML uses bare
= (oper: =) which trips PyYAML's value tag — the
scan registers a constructor that reads it as a string.
- System / sample Liveboards are export-FORBIDDEN for a normal identity; the scan
counts them separately ("system/locked") rather than treating that as failure.
- Privacy: see
PRIVACY.md. Output shapes: see refs/output-shapes.md. Readout
layout: see refs/readout-template.md.
