Brand Vocabulary Check tool
The Brand Vocabulary Check tool scans each block's source text for vocabulary violations defined by a brand voice profile. Forbidden terms are flagged at major severity, competitor terms at critical severity, and where a forbidden term has a preferred replacement, the finding includes that suggestion. When a termbase is provided, the tool also looks up brand-vocabulary terms in it, flagging competitor and forbidden entries it finds. Matches are case-insensitive.
Findings are recorded on the block's properties and an overall brand-voice score is attached as an annotation. This is a rule-based check that needs no LLM credentials, and it is designed to run before the LLM-based brand voice check. The voice profile and termbase are supplied programmatically rather than as schema parameters; the profile can also be resolved lazily from an organizational context hierarchy.
This tool has no configurable parameters.
Examples
Add to a flow
The voice profile and termbase are supplied programmatically; reference the tool with an empty configuration.
{}Processing notes
Operates on blocks; blocks with empty source text are skipped.
Without a resolved profile or termbase, no findings are produced.
Findings are stored on the block and a brand-voice annotation with an overall score is attached.
Limitations
The voice profile and termbase are configured programmatically; they are not exposed as schema parameters.
Term matching is case-insensitive substring matching; it does not account for inflection or morphology.
Evaluates source text.
← Back to the Tool Reference