Skip to content
OnFire OnFire Docs

ai_prospecting_field_glossary

Returns the canonical JSON contract for every field ai_prospecting can produce: type, expected values, what each field means, and (for high-signal fields) how to use it.

When to use it

Section titled “When to use it”
  • Once per conversation, before or right after the first ai_prospecting run, so the agent can interpret every column in top_picks / preview_rows and in rows pulled from the run’s dataset via query_datasets.
  • Only if your MCP client does not auto-load resources. Clients that support MCP resource injection already load the same content from onfire://ai_prospecting/field_glossary at session start.

Inputs

Section titled “Inputs”

None.

Returns

Section titled “Returns”

JSON. Top-level keys are field names (e.g. MASTER_SCORE_REASON, SCORE_WARM_INTRO, WORKED_IN_CLIENT_COMPANY_IN_PAST). Each value is an object with type, what_it_means, and — for high-signal fields — how_to_use plus examples / expected_values. Self-referential entries cover response-level meta fields (field_index, field_glossary) so the agent understands what it just received.

Notes

Section titled “Notes”
  • The glossary is static metadata, not per-tenant data, and is byte-stable across calls within a server version. Re-fetching within the same conversation is pure waste.
  • Same content as the onfire://ai_prospecting/field_glossary MCP resource — pick whichever path your client supports.