get_qualified_search_result

The get_qualified_search_result tool reads a qualified search started with search_qualified_talents_by_prompt or search_qualified_talents_by_filters. It only reads talents that already received a verdict and never costs credits. Talents are returned green first, then amber, then red, in judgement order within a verdict. Calls, waves, and credits are explained on the qualified search overview. Follow nextAction:

`nextAction`	What the assistant should do
`wait`	A call is running. Call this tool again in 10–20 seconds.
`continue`	The call stopped and more matching talents exist. Call `continue_qualified_search` to authorise another call. It looks for `target.greens` new greens, up to `target.maxCredits`.
`retry`	The last wave failed. Call `continue_qualified_search` to retry it.
`done`	The talent pool is exhausted. Stop.

canContinue is kept for compatibility and equals nextAction === "continue". The HTTP counterpart is GET /v1/search/talents/qualified/{castingId}.

Input

Parameter	Type	Required	Description
`castingId`	string	Yes	Identifier returned when the qualified search was started.
`verdicts`	string[]	No	One or more of `green`, `amber`, `red`. Default: all three. Filters only the returned talents, never `counts` or credits.
`cursor`	string	No	`nextPageCursor` from a previous call. Omit it on the first page. Bound to the `verdicts` it was issued with.
`pageSize`	number	No	Number of talents to return, from 1 to 100. Default 50.

Response

The assistant receives a JSON payload with talents and a short summary table. The structured result contains:

Field	Type	Description
`castingId`	string	Identifier of the qualified search
`status`	string	`running`, `can_continue`, `pool_exhausted`, or `failed_retryable`
`nextAction`	string	`wait`, `continue`, `retry`, or `done`
`canContinue`	boolean	`true` only when `nextAction` is `continue`
`scoringLanguage`	string	`en` or `fr`: language of `recap`, criterion `title`, and criterion `result` for every talent of this search. Fixed when the search started. `null` for searches started before this field existed.
`counts`	object	`{ green, amber, red }` for the whole search, regardless of `verdicts`
`creditsCharged`	number	Credits charged on this search so far, across all calls
`creditsRemaining`	number	Credits the current call can still spend: the remaining call budget, bounded by what the workspace can pay (credits reserved for this call plus the available balance). `0` once the call is over. Can go up or down while the call runs because the balance is shared; read `creditsCharged` for what was spent. With `nextAction` `retry` and an empty balance it is `0`, and `continue_qualified_search` reports insufficient credits until credits are added
`nextPageCursor`	string	Next page cursor, or `null`

The JSON text also includes talents. Each talent has talentId, aiVerdict (green, amber, or red), scoring (evaluation result or null: recap and criterias, each with title, result, status, evaluable, references; recap, title, and result are written in scoringLanguage, quotes in references stay in the profile’s original language), and profile (same profile fields as other Kalent talent search tools). Talents that cannot be loaded are omitted.

Errors

The tool response is marked as an error when there is no MCP session, the qualified search is not found (Casting not found.), the cursor does not match this search and verdict filter, or loading results fails.

Getting Started

Embedded

Platform

get_qualified_search_result

Input

Response

Errors

​Input

​Response

​Errors

Input

Response

Errors