Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.speckle.systems/llms.txt

Use this file to discover all available pages before exploring further.

​
In Intelligence Dashboards

This page covers Data Validation (checks and standards) detail and history views. For dashboard-based validation widget result reading, see Validation Widgets in Dashboards.

​
What you see on the page

There are two high-level result views used for quick status checks:
  • Project page Data Validation widget: compact check cards in the project page for quick scanning.
  • Checks page cards: check cards with score, status, trend hint, and scoped model count.

​
Practical review flow

These high-level views are a KPI heads-up layer for rules defined by your team. Use them to spot drift quickly, then drill into a specific check for detailed analysis. When you open a check result page, you will typically see:

​
Result states

  • a top-level check score and status
  • a per-rule list showing pass/warn/fail outcomes
  • expandable rule rows to inspect logic flow (WHERE -> CHECK) and counts at each step
  • model cards for each model included in the check scope
  • latest run details (time, model/version context)
  • historical trend for previous runs
  • object-level detail table for investigation, including passing and failing objects | Status | Meaning | | --------- | --------------------------------------------------------- | | PASS | Pass-rate ≥ configured pass threshold. | | WARN | Between warn/pass band when configured. | | FAIL | Below thresholds. | | PENDING | Evaluation still running in background for that rollup across models, versions, and rules. |
This top section is the aggregate result for the check across all models in scope. Selecting View result from a model card opens that model’s detailed result, where you can inspect rule-by-rule breakdown, the 3D viewer state, and the data table. Threshold tuning: Checks — Thresholds and status.

​
If a result looks unexpected, check:

1

Open latest check result

Start with current check score and top failing rules.
2

Drill into failing rules

Expand a rule to inspect its logic flow (WHERE and CHECK) and step-by-step counts. Rule-level viewing also populates the object table with pass/fail rows so you can inspect individual objects directly.
3

Compare with history

Confirm whether failures are regressions, recurring debt, or resolved trends.
4

Share and coordinate

If external coordination tooling is used, include BCF export in handoff.
  1. Same model/version you expected auditing?
  2. Predicates/threshold edits since that baseline snapshot?
  3. Failures clustered by stream or timeframe (data drift) vs sporadic flake?

​
Model and version context

Plans cap how many models attach. Every row references explicit version ids plus trigger + timestamp—carry that along when debating trends.

​
Coordination handoff

BCF export from validation emits topic scaffolding plus failures so coordinators triage outside Speckle.

​
FAQ

PENDING appears when no completed evaluation exists yet for that rule/run context.
Yes. Warning and pass thresholds can be configured globally and per rule in authoring flow.
Use both. Latest helps with immediate action; history helps with governance and trend confidence.
Last modified on May 11, 2026