describe_onfire_schema

describe_onfire_schema is the authoritative answer to “does this field exist on this entity, and can I filter on it”. It returns the semantic model behind ask_onfire — the shape you author a QueryIR against.

When to use it

• Before querying an entity whose fields you haven’t seen. Batch every such entity into one call.
• After a FIELD_NOT_FOUND validation error, to get the exact names.
• To list the queryable entities when the inlined catalog isn’t available to your client (show_catalog=True).

Inputs

Parameter	Type	Required	Description
ids	string[]	No	Entity identifiers, e.g. ["contact", "company"]. Omit for the catalog.
show_catalog	boolean	No	Return the routing catalog as text instead of JSON. Default false.

Returns

• With ids — {"schemas": {id: {...}}}: the full schema per entity.
• With show_catalog=True — the routing catalog as text.
• With no arguments — the same catalog as JSON ({"catalog": [...]}).
• Unknown identifier — {"error", "valid", "did_you_mean"}.

Each entity schema separates three field kinds:

Kind	Filterable	Selectable	Example
Dimension	Yes (when flagged filterable)	Yes	location_country, size_band, linkedin_url
Attribute	No	Yes	full_name, summary, start_date
Measure	n/a	Yes	Pre-aggregated counts

Dimensions carry closed-list values where they have them. The schema also carries the entity’s insight-search capability, joins, worked examples, and per-entity rules.

Notes & limits

• Filter only on dimensions flagged filterable, or via insight_filters. Filtering an attribute is rejected at validation time.
• The catalog is loaded from the warehouse and refreshed periodically, so new entities appear without a server deploy — treat this tool, not any static list, as the source of truth.
• This is the only data-query surface: there is no raw-SQL tool against the data lake. (SQL over your own saved results is query_datasets.)
• Read-only, free, and static within a server version.