Use existing account search for saved companies and discovery preview for new companies. Inspect actor-selected ICP rubrics and account scope before grading/research; explain the quote and obtain exact authorization. Poll job/discovery status. Analytics reports describe recorded evidence, not causal attribution or guaranteed historical coverage.
Read an account's buying committee
accounts.buyer_groups.get.v2
Use when: See who is on an account and what buying role each holds, ordered by influence.
Avoid when: Do not use this to search people across the workspace — that is contacts.search.v2.
Effects: Reads current workspace state.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.buyer_groups.get.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"account_id": {
"description": "Exact account UUID returned by a Hivekind search, list, get, preview, or resource call.",
"format": "uuid",
"title": "Account Id",
"type": "string"
}
},
"required": [
"account_id"
],
"title": "McpPublicAccountsBuyerGroupsGetInputV2",
"type": "object"
}
Use when: Place an account's existing contacts into buying roles. Declines when the account has too few contacts to be a committee.
Avoid when: Do not use this to find people you do not have; it only organises the contacts already on the account.
Effects: Queues classification and meters one unit per contact classified. Poll jobs.status.get.v2.
Cost: buyer_classification Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.classify_buyers.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"action_proposal_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional. Hivekind fills this from the MCP consent grant.",
"title": "Action Proposal Id"
},
"account_id": {
"description": "Exact account UUID returned by a Hivekind search, list, get, preview, or resource call.",
"format": "uuid",
"title": "Account Id",
"type": "string"
},
"methodology": {
"anyOf": [
{
"maxLength": 64,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Sales methodology whose role list to classify against. Omit for the workspace default.",
"title": "Methodology"
},
"min_contacts": {
"default": 2,
"description": "Do nothing unless the account has at least this many contacts. One contact is not a committee — labelling a solo founder the economic buyer costs a credit and says nothing. Pass 1 to classify regardless.",
"maximum": 1000,
"minimum": 1,
"title": "Min Contacts",
"type": "integer"
},
"include_classified": {
"default": false,
"description": "By default only contacts with no committed buyer_role are sent, so re-running is cheap. Set true to reclassify everyone.",
"title": "Include Classified",
"type": "boolean"
}
},
"required": [
"account_id"
],
"title": "McpMcpPublicAccountsClassifyBuyersInputV2",
"type": "object"
}
Use when: You want new companies outside the workspace and need the exact rubric, order size, credit quote, and conflict state before starting.
Avoid when: You are searching accounts already in Hivekind; use accounts.search.v2 instead.
Effects: No write, provider query, grading, account creation, or charge.
Cost: Free. Returns the exact maximum delivery charge. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.discovery.preview.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"count": {
"default": 25,
"description": "How many accounts to deliver. This is an order size, not a search limit: raising it grades proportionally more companies and costs proportionally more.",
"maximum": 100,
"minimum": 1,
"title": "Count",
"type": "integer"
},
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Rubric to discover against. Omit for the caller's active one.",
"title": "Rubric Id"
},
"owner_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Who the delivered accounts are assigned to. Omit to let it resolve: to you when you grade against this rubric, otherwise to its sole holder. An account is graded against its owner's rubric, so an owner who does not hold this one is refused, as is a rubric nobody holds — preview it to see the resolved owner, or the reason there is none.",
"title": "Owner Id"
},
"recompile": {
"default": false,
"description": "Re-author the Clay query instead of reusing the pinned one. Authoring is an LLM step and is not stable, so the new query will not be the old query. Use this when the rubric itself has changed — not to retry a week that returned few accounts.",
"title": "Recompile",
"type": "boolean"
}
},
"title": "McpAccountsDiscoveryPreviewInputV2",
"type": "object"
}
Use when: You approved the exact order returned by accounts.discovery.preview.v2 and want to discover, grade, and add new matching companies.
Avoid when: You have not previewed the order, need existing accounts, or another run for the rubric is active.
Effects: Queues a background provider workflow and adds delivered accounts. Poll accounts.discovery.status.get.v2.
Cost: Billed per account delivered at the previewed rate; failed and undelivered accounts are free. Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.discovery.start.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"count": {
"default": 25,
"description": "How many accounts to deliver. This is an order size, not a search limit: raising it grades proportionally more companies and costs proportionally more.",
"maximum": 100,
"minimum": 1,
"title": "Count",
"type": "integer"
},
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Rubric to discover against. Omit for the caller's active one.",
"title": "Rubric Id"
},
"owner_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Who the delivered accounts are assigned to. Omit to let it resolve: to you when you grade against this rubric, otherwise to its sole holder. An account is graded against its owner's rubric, so an owner who does not hold this one is refused, as is a rubric nobody holds — preview it to see the resolved owner, or the reason there is none.",
"title": "Owner Id"
},
"recompile": {
"default": false,
"description": "Re-author the Clay query instead of reusing the pinned one. Authoring is an LLM step and is not stable, so the new query will not be the old query. Use this when the rubric itself has changed — not to retry a week that returned few accounts.",
"title": "Recompile",
"type": "boolean"
},
"action_proposal_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional. Hivekind fills this from the MCP consent grant.",
"title": "Action Proposal Id"
}
},
"title": "McpMcpAccountsDiscoveryStartInputV2",
"type": "object"
}
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.discovery.status.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"run_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Omit for the most recent run in this workspace.",
"title": "Run Id"
},
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Narrows the 'most recent run' lookup to one rubric. Ignored when run_id is given.",
"title": "Rubric Id"
}
},
"title": "McpAccountsDiscoveryStatusInputV2",
"type": "object"
}
Access: accounts:read, research:run. Approval: operation authorization. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "accounts.research.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001",
"name": "Example name",
"objective": "Example objective"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"account_id": {
"description": "Exact account UUID returned by a Hivekind search, list, get, preview, or resource call.",
"format": "uuid",
"title": "Account Id",
"type": "string"
},
"name": {
"description": "Human-readable name for the record.",
"maxLength": 255,
"minLength": 1,
"title": "Name",
"type": "string"
},
"domain": {
"anyOf": [
{
"maxLength": 255,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Domain used by the public accounts research input v2 operation.",
"title": "Domain"
},
"objective": {
"description": "Plain-language outcome the user wants from this operation.",
"maxLength": 2000,
"minLength": 1,
"title": "Objective",
"type": "string"
}
},
"required": [
"account_id",
"name",
"objective"
],
"title": "McpPublicAccountsResearchInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: Campaign and loop execution metrics. Loop execution counts; no program spend or causal ROI.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.campaigns.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"time_period": {
"default": "quarter",
"description": "Time period used by the analytics campaigns get input v2 operation.",
"enum": [
"day",
"week",
"month",
"quarter"
],
"title": "Time Period",
"type": "string"
}
},
"title": "McpAnalyticsCampaignsGetInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: Recorded outcome counts. Recorded outcomes and unmatched records; no causal attribution or commercial commitments.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.outcomes.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Revenue outcome overview. Aggregate KPIs; not a historical rep-by-tier cohort funnel.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.overview.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Pipeline snapshot comparison. Compare available snapshots; not eight-quarter win/loss cohorts or a commitment baseline.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.pipeline.compare.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Available pipeline snapshot dates. Available snapshots and periods only; resolve exact comparison dates before requesting historical analysis.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.pipeline.dates.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Pre-pipeline funnel and owner coverage. Current readiness and owner scorecard; not historical owner/tier conversion.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.readiness.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"owner_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Existing workspace member UUID; omitted means workspace scope.",
"title": "Owner Id"
},
"owner_scope": {
"default": "any",
"description": "Use mine for the authenticated actor; do not guess or send the actor identity. Cannot combine mine with owner_id.",
"enum": [
"any",
"mine"
],
"title": "Owner Scope",
"type": "string"
}
},
"title": "McpPublicAnalyticsReadinessGetInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: LinkedIn individual and team match coverage. Consented LinkedIn match coverage only; not board/investor/advisor paths, relationship strength, or proof that an unmatched account is cold.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.relationships.coverage.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Signal geography and coverage. Recorded signals by geography; not collector completeness, urgency scoring, or a notification subscription.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.signals.geography.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"start_date": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Start date used by the analytics signals geography get input v2 operation.",
"title": "Start Date"
},
"end_date": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "End date used by the analytics signals geography get input v2 operation.",
"title": "End Date"
},
"owner_id": {
"default": [],
"description": "Exact owner UUID returned by a Hivekind search, list, get, preview, or resource call.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 50,
"title": "Owner Id",
"type": "array"
},
"region": {
"default": [],
"description": "Region used by the analytics signals geography get input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Region",
"type": "array"
},
"signal_type": {
"default": [],
"description": "Signal type used by the analytics signals geography get input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Signal Type",
"type": "array"
},
"icp_grade": {
"default": [],
"description": "Icp grade used by the analytics signals geography get input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 8,
"title": "Icp Grade",
"type": "array"
},
"granularity": {
"default": "region",
"description": "Granularity used by the analytics signals geography get input v2 operation.",
"enum": [
"city",
"state",
"country",
"region"
],
"title": "Granularity",
"type": "string"
},
"signal_scope": {
"default": "important",
"description": "Signal scope used by the analytics signals geography get input v2 operation.",
"enum": [
"important",
"all"
],
"title": "Signal Scope",
"type": "string"
},
"owner_scope": {
"default": "any",
"description": "Use mine for the authenticated actor; do not guess or send the actor identity. Cannot combine mine with owner_id.",
"enum": [
"any",
"mine"
],
"title": "Owner Scope",
"type": "string"
}
},
"title": "McpAnalyticsSignalsGeographyGetInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: Tier and ICP account coverage. Current tier/grade matrix and paginated list; no historical tier reassignment.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "analytics.tiers.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"owner_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Existing workspace member UUID; omitted means workspace scope.",
"title": "Owner Id"
},
"tier": {
"anyOf": [
{
"maxLength": 64,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Tier used by the public analytics tiers get input v2 operation.",
"title": "Tier"
},
"icp_grade": {
"anyOf": [
{
"maxLength": 32,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Icp grade used by the public analytics tiers get input v2 operation.",
"title": "Icp Grade"
},
"industry": {
"anyOf": [
{
"maxLength": 255,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Industry used by the public analytics tiers get input v2 operation.",
"title": "Industry"
},
"limit": {
"default": 100,
"description": "Maximum number of records to return in this page.",
"maximum": 500,
"minimum": 1,
"title": "Limit",
"type": "integer"
},
"offset": {
"default": 0,
"description": "Zero-based number of matching records to skip for pagination.",
"maximum": 1000000,
"minimum": 0,
"title": "Offset",
"type": "integer"
},
"view": {
"default": "accounts",
"description": "View used by the public analytics tiers get input v2 operation.",
"enum": [
"accounts",
"contacts"
],
"title": "View",
"type": "string"
},
"owner_scope": {
"default": "any",
"description": "Use mine for the authenticated actor; do not guess or send the actor identity. Cannot combine mine with owner_id.",
"enum": [
"any",
"mine"
],
"title": "Owner Scope",
"type": "string"
}
},
"title": "McpPublicAnalyticsTiersGetInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: Account score evidence and rubric. Existing score breakdown and the rubric actually used, not new evaluation.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.account.breakdown.get.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Account score and rubric version history. Existing per-run score history; not historical win/loss cohort attribution.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.account.history.get.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Propose client-fit rubric changes. Model proposal only; does not persist or rescore. Not a historical holdout experiment.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Model proposal only; does not persist or rescore. Not a historical holdout experiment.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.clients.suggest.v2",
"arguments": {
"rubric_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Client fit validation. Current/prior clients owned by rubric assignees; not closed-won/lost deal history.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.clients.validation.get.v2",
"arguments": {
"rubric_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Propose CRM-evidence rubric dimensions. Requires connected CRM and built won-customer profile. Does not persist or claim signal causality.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Requires connected CRM and built won-customer profile. Does not persist or claim signal causality.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.crm.suggest.v2",
"arguments": {
"rubric_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Read the workspace ICP rubric the user already approved.
Avoid when: Do not use this to score an account — that is icp.score.get.v2.
Effects: Reads current workspace state.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: icp:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.get.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Count what a grading scope would cover and what it would cost, before spending anything.
Avoid when: Do not use this to grade — that is icp.grading.start.v2.
Effects: Writes nothing and grades nothing. Deliberately not credit-gated: a workspace out of credits is the one that needs the number.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: scoring:run. Approval: operation authorization. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.grading.preview.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Grade against this rubric and everyone assigned to it. Omit to use the calling user's own pinned rubric and only their accounts. Setting it does not change what any user has pinned.",
"title": "Rubric Id"
},
"account_ids": {
"default": [],
"description": "Grade exactly these accounts. Resolve the request to a concrete set first — accounts.search.v2, an event roster, or the account_ids an import returns — then pass them here. This is the right answer whenever the user means a particular set rather than a rule. It narrows: archived and scoring-exempt accounts stay excluded even if listed.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 500,
"title": "Account Ids",
"type": "array"
},
"only_unscored": {
"default": false,
"description": "Skip accounts already scored under the TARGET RUBRIC. Rubric-relative and workspace-wide, so after importing a handful of accounts this still selects every other account nobody has scored under that rubric. For 'what I just added', pass account_ids.",
"title": "Only Unscored",
"type": "boolean"
},
"never_scored": {
"default": false,
"description": "No grade under any rubric. Usually what 'unscored' means.",
"title": "Never Scored",
"type": "boolean"
},
"industries": {
"default": [],
"description": "Industries used by the icp grading preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Industries",
"type": "array"
},
"regions": {
"default": [],
"description": "Regions used by the icp grading preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Regions",
"type": "array"
},
"account_types": {
"default": [],
"description": "Prospect, Client, Partner and so on. Competitor and Self are never scored, so asking for them returns nothing.",
"items": {
"type": "string"
},
"maxItems": 16,
"title": "Account Types",
"type": "array"
},
"icp_grades": {
"default": [],
"description": "Icp grades used by the icp grading preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 8,
"title": "Icp Grades",
"type": "array"
},
"owner_ids": {
"default": [],
"description": "Account owners. Use icp.scope.facets.v2 to resolve names to ids.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 50,
"title": "Owner Ids",
"type": "array"
},
"minimum_icp_score": {
"anyOf": [
{
"maximum": 100,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"description": "Minimum icp score used to bound matching records.",
"title": "Minimum Icp Score"
},
"maximum_icp_score": {
"anyOf": [
{
"maximum": 100,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"description": "Maximum icp score used to bound matching records.",
"title": "Maximum Icp Score"
},
"minimum_employee_count": {
"anyOf": [
{
"maximum": 10000000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Minimum employee count used to bound matching records.",
"title": "Minimum Employee Count"
},
"maximum_employee_count": {
"anyOf": [
{
"maximum": 10000000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Maximum employee count used to bound matching records.",
"title": "Maximum Employee Count"
},
"contact_count_min": {
"anyOf": [
{
"maximum": 100000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Contact count min used by the icp grading preview input v2 operation.",
"title": "Contact Count Min"
},
"contact_count_max": {
"anyOf": [
{
"maximum": 100000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Contact count max used by the icp grading preview input v2 operation.",
"title": "Contact Count Max"
},
"created_since": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Accounts added on or after this instant.",
"title": "Created Since"
},
"created_before": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Created before used by the icp grading preview input v2 operation.",
"title": "Created Before"
},
"not_scored_since": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Only accounts with no score recorded since this instant. Never-scored accounts count as stale.",
"title": "Not Scored Since"
}
},
"title": "McpIcpGradingPreviewInputV2",
"type": "object"
}
Output schema inside the result envelope
{
"additionalProperties": false,
"properties": {
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"title": "Rubric Id"
},
"total_in_scope": {
"default": 0,
"minimum": 0,
"title": "Total In Scope",
"type": "integer"
},
"would_score": {
"default": 0,
"description": "What the run will actually grade.",
"minimum": 0,
"title": "Would Score",
"type": "integer"
},
"already_scored": {
"default": 0,
"minimum": 0,
"title": "Already Scored",
"type": "integer"
},
"skipped_no_rubric": {
"default": 0,
"description": "Matched the filters but their owner has no rubric pinned, so the dispatcher would drop them.",
"minimum": 0,
"title": "Skipped No Rubric",
"type": "integer"
},
"credits_per_account": {
"default": 0,
"minimum": 0,
"title": "Credits Per Account",
"type": "number"
},
"estimated_credits": {
"default": 0,
"minimum": 0,
"title": "Estimated Credits",
"type": "number"
},
"credit_balance": {
"anyOf": [
{
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"title": "Credit Balance"
},
"sufficient_credits": {
"default": true,
"title": "Sufficient Credits",
"type": "boolean"
},
"requested_account_ids": {
"default": 0,
"description": "How many ids the caller asked for.",
"minimum": 0,
"title": "Requested Account Ids",
"type": "integer"
},
"excluded_not_active": {
"default": 0,
"description": "Requested but archived or candidate.",
"minimum": 0,
"title": "Excluded Not Active",
"type": "integer"
},
"excluded_exempt_type": {
"default": 0,
"description": "Requested but Competitor, Self or Partner.",
"minimum": 0,
"title": "Excluded Exempt Type",
"type": "integer"
},
"excluded_other_owner": {
"default": 0,
"description": "Requested but owned by someone outside the rubric's assignees. Name that owner in owner_ids, or grade against their rubric.",
"minimum": 0,
"title": "Excluded Other Owner",
"type": "integer"
},
"excluded_not_found": {
"default": 0,
"description": "Requested but not an account in this workspace.",
"minimum": 0,
"title": "Excluded Not Found",
"type": "integer"
}
},
"title": "IcpGradingPreviewResultV2",
"type": "object"
}
Start ICP grading
icp.grading.start.v2
Use when: You approved the exact account scope and cost returned by icp.grading.preview.v2 and want fresh asynchronous grading.
Avoid when: You only need existing audited scores; use icp.scores.get.v2.
Effects: Queues metered grading. Poll jobs.status.get.v2 with the returned job_id.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.grading.start.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Grade against this rubric and everyone assigned to it. Omit to use the calling user's own pinned rubric and only their accounts. Setting it does not change what any user has pinned.",
"title": "Rubric Id"
},
"account_ids": {
"default": [],
"description": "Grade exactly these accounts. Resolve the request to a concrete set first — accounts.search.v2, an event roster, or the account_ids an import returns — then pass them here. This is the right answer whenever the user means a particular set rather than a rule. It narrows: archived and scoring-exempt accounts stay excluded even if listed.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 500,
"title": "Account Ids",
"type": "array"
},
"only_unscored": {
"default": false,
"description": "Skip accounts already scored under the TARGET RUBRIC. Rubric-relative and workspace-wide, so after importing a handful of accounts this still selects every other account nobody has scored under that rubric. For 'what I just added', pass account_ids.",
"title": "Only Unscored",
"type": "boolean"
},
"never_scored": {
"default": false,
"description": "No grade under any rubric. Usually what 'unscored' means.",
"title": "Never Scored",
"type": "boolean"
},
"industries": {
"default": [],
"description": "Industries used by the icp grading start input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Industries",
"type": "array"
},
"regions": {
"default": [],
"description": "Regions used by the icp grading start input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Regions",
"type": "array"
},
"account_types": {
"default": [],
"description": "Prospect, Client, Partner and so on. Competitor and Self are never scored, so asking for them returns nothing.",
"items": {
"type": "string"
},
"maxItems": 16,
"title": "Account Types",
"type": "array"
},
"icp_grades": {
"default": [],
"description": "Icp grades used by the icp grading start input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 8,
"title": "Icp Grades",
"type": "array"
},
"owner_ids": {
"default": [],
"description": "Account owners. Use icp.scope.facets.v2 to resolve names to ids.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 50,
"title": "Owner Ids",
"type": "array"
},
"minimum_icp_score": {
"anyOf": [
{
"maximum": 100,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"description": "Minimum icp score used to bound matching records.",
"title": "Minimum Icp Score"
},
"maximum_icp_score": {
"anyOf": [
{
"maximum": 100,
"minimum": 0,
"type": "number"
},
{
"type": "null"
}
],
"default": null,
"description": "Maximum icp score used to bound matching records.",
"title": "Maximum Icp Score"
},
"minimum_employee_count": {
"anyOf": [
{
"maximum": 10000000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Minimum employee count used to bound matching records.",
"title": "Minimum Employee Count"
},
"maximum_employee_count": {
"anyOf": [
{
"maximum": 10000000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Maximum employee count used to bound matching records.",
"title": "Maximum Employee Count"
},
"contact_count_min": {
"anyOf": [
{
"maximum": 100000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Contact count min used by the icp grading start input v2 operation.",
"title": "Contact Count Min"
},
"contact_count_max": {
"anyOf": [
{
"maximum": 100000,
"minimum": 0,
"type": "integer"
},
{
"type": "null"
}
],
"default": null,
"description": "Contact count max used by the icp grading start input v2 operation.",
"title": "Contact Count Max"
},
"created_since": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Accounts added on or after this instant.",
"title": "Created Since"
},
"created_before": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Created before used by the icp grading start input v2 operation.",
"title": "Created Before"
},
"not_scored_since": {
"anyOf": [
{
"format": "date-time",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Only accounts with no score recorded since this instant. Never-scored accounts count as stale.",
"title": "Not Scored Since"
},
"action_proposal_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Optional. Hivekind fills this from the MCP consent grant.",
"title": "Action Proposal Id"
}
},
"title": "McpMcpIcpGradingStartInputV2",
"type": "object"
}
Use when: Use the existing Core workflow: Full grading rubric. Existing workspace rubric and provenance; read only.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.rubric.details.get.v2",
"arguments": {
"rubric_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: Use the existing Core workflow: Select active grading rubric. Changes only the calling user’s active rubric; does not score accounts.
Avoid when: Do not infer unsupported history, attribution, or commitments from this response.
Effects: Changes only the calling user’s active rubric; does not score accounts.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: actions:execute. Approval: exact_scope_single_action. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.rubric.pin.v2",
"arguments": {
"rubric_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: icp:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.rubrics.list.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"scope": {
"default": "user",
"description": "Assigned rubrics or all rubrics visible in this workspace.",
"enum": [
"user",
"workspace"
],
"title": "Scope",
"type": "string"
},
"limit": {
"default": 25,
"description": "Maximum number of records to return in this page.",
"maximum": 100,
"minimum": 1,
"title": "Limit",
"type": "integer"
}
},
"title": "McpIcpRubricsListInputV2",
"type": "object"
}
Use when: List the industries, owners and grades that actually have accounts behind them, so a grading scope cannot be built from values that match nothing.
Avoid when: Do not use this to count what a scope would grade — that is icp.grading.preview.v2.
Effects: Reads current workspace state.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: icp:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.scope.facets.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"rubric_id": {
"anyOf": [
{
"format": "uuid",
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Exact rubric UUID returned by a Hivekind search, list, get, preview, or resource call.",
"title": "Rubric Id"
},
"only_unscored": {
"default": false,
"description": "Only unscored used by the public icp scope facets input v2 operation.",
"title": "Only Unscored",
"type": "boolean"
}
},
"title": "McpPublicIcpScopeFacetsInputV2",
"type": "object"
}
Use when: Read the latest audited score for your active rubric, with no new scoring or charge. A matched rubric disqualifier pins the grade to the lowest tier whatever the score, reported as disqualified and disqualification_reason. Use icp.grading.preview.v2 followed by icp.grading.start.v2 for fresh evaluation.
Avoid when: Do not use this to read the rubric — that is icp.get.v2.
Effects: Reads current workspace state.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read, icp:read, scoring:run. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.score.get.v2",
"arguments": {
"account_id": "00000000-0000-4000-8000-000000000001"
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: You have exact account IDs and need audited scores for the signed-in actor's active rubric.
Avoid when: You need fresh grading; use the Intelligence grading preview and start tools.
Effects: Returns scored, unscored, or no-active-rubric state per account. Read only.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: accounts:read, icp:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "icp.scores.get.v2",
"arguments": {
"account_ids": [
"00000000-0000-4000-8000-000000000001"
]
},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Use when: CRO snapshot of pipeline stages and counts.
Avoid when: Do not use this to list campaigns — that is campaigns.list.v2.
Effects: Reads current workspace state.
Cost: No gateway credit charge; reads saved workspace data without starting paid provider work. Gateway rate code: none; downstream billing may apply.
Access: runs:read. Approval: operation authorization. Reads do not need an operation UUID.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "pipeline.summary.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"lookback_days": {
"default": 90,
"description": "Number of calendar days before today to include.",
"maximum": 730,
"minimum": 1,
"title": "Lookback Days",
"type": "integer"
},
"maximum_opportunities": {
"default": 1000,
"description": "Maximum opportunities used to bound matching records.",
"maximum": 5000,
"minimum": 1,
"title": "Maximum Opportunities",
"type": "integer"
}
},
"title": "McpPublicPipelineSummaryInputV2",
"type": "object"
}
Use when: You need accounts with verified LinkedIn paths ranked by saved ICP grade/score, including connected ungraded accounts.
Avoid when: You need one known account or only account-level filters; use accounts.search.v2.
Effects: No-write preview. Filters consented path eligibility before ranking and returns graded and ungraded sections separately.
Cost: No gateway rate is configured. Check the Core workflow preview/quote for downstream charges and final settlement. Gateway rate code: none; downstream billing may apply.
Access: accounts:read. Approval: operation authorization. A stable operation UUID is required.
Keep the same operation and operation_id when retrying identical work. Reconcile an uncertain result before new work; never replay an old V1 write under a new V2 ID.
Check status, execution_state, error_code, output and evidence/resource references. Queued work is not complete: poll jobs.status.get.v2 in domain.workspace.v2. For read-back verification require read_back_verified=true where returned; reconcile failures without creating another object.
Invocation example using fictional fixtures
{
"tool": "domain.intelligence.v2",
"arguments": {
"params": {
"action": "invoke",
"operation": "targets.preview.v2",
"arguments": {},
"operation_id": "00000000-0000-4000-8000-000000000003"
}
},
"notice": "Fictional fixture IDs and example text. Resolve real IDs, use returned preview tokens and approve exact side effects before execution. Input validation is not proof of provider readiness or a successful business outcome."
}
Input schema: fields, defaults and constraints
{
"additionalProperties": false,
"properties": {
"query": {
"default": "*",
"description": "Optional account name, domain, or industry text; * means any account.",
"maxLength": 255,
"minLength": 1,
"title": "Query",
"type": "string"
},
"account_ids": {
"default": [],
"description": "Optional exact account UUID scope resolved from a prior search.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 500,
"title": "Account Ids",
"type": "array"
},
"owner_scope": {
"default": "workspace",
"description": "Account ownership scope; mine is derived from the signed-in actor.",
"enum": [
"mine",
"workspace",
"assigned",
"unassigned"
],
"title": "Owner Scope",
"type": "string"
},
"owner_ids": {
"default": [],
"description": "Optional exact owner UUIDs; cannot be combined with mine or unassigned.",
"items": {
"format": "uuid",
"type": "string"
},
"maxItems": 100,
"title": "Owner Ids",
"type": "array"
},
"industries": {
"default": [],
"description": "Industries used by the targets preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Industries",
"type": "array"
},
"regions": {
"default": [],
"description": "Regions used by the targets preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 50,
"title": "Regions",
"type": "array"
},
"account_types": {
"default": [],
"description": "Account types used by the targets preview input v2 operation.",
"items": {
"type": "string"
},
"maxItems": 20,
"title": "Account Types",
"type": "array"
},
"buyer_roles": {
"default": [],
"description": "Require at least one active account contact with one of these buyer roles.",
"items": {
"type": "string"
},
"maxItems": 20,
"title": "Buyer Roles",
"type": "array"
},
"signal_types": {
"default": [],
"description": "Require at least one recorded account/contact signal of one of these types.",
"items": {
"type": "string"
},
"maxItems": 20,
"title": "Signal Types",
"type": "array"
},
"signal_lookback_days": {
"default": 30,
"description": "Signal lookback days used by the targets preview input v2 operation.",
"maximum": 365,
"minimum": 1,
"title": "Signal Lookback Days",
"type": "integer"
},
"network_scope": {
"default": "workspace_consented",
"description": "Whether verified paths are limited to self or may include sharing teammates.",
"enum": [
"self",
"workspace_consented"
],
"title": "Network Scope",
"type": "string"
},
"require_verified_path": {
"default": true,
"description": "When true, path eligibility is evaluated before ranking and pagination.",
"title": "Require Verified Path",
"type": "boolean"
},
"icp_score_state": {
"default": "any",
"description": "Icp score state used by the targets preview input v2 operation.",
"enum": [
"any",
"scored_only",
"unscored_only"
],
"title": "Icp Score State",
"type": "string"
},
"icp_grades": {
"default": [],
"description": "Icp grades used by the targets preview input v2 operation.",
"items": {
"enum": [
"A",
"B",
"C",
"D",
"E",
"F"
],
"type": "string"
},
"maxItems": 6,
"title": "Icp Grades",
"type": "array"
},
"include_ungraded": {
"default": true,
"description": "Return path-qualified accounts without a complete saved grade/score separately.",
"title": "Include Ungraded",
"type": "boolean"
},
"ranking": {
"default": "grade_then_score",
"description": "Ranking used by the targets preview input v2 operation.",
"enum": [
"grade_then_score",
"score_desc",
"signal_then_score"
],
"title": "Ranking",
"type": "string"
},
"graded_limit": {
"default": 25,
"description": "Graded limit used by the targets preview input v2 operation.",
"maximum": 100,
"minimum": 1,
"title": "Graded Limit",
"type": "integer"
},
"ungraded_limit": {
"default": 100,
"description": "Ungraded limit used by the targets preview input v2 operation.",
"maximum": 100,
"minimum": 1,
"title": "Ungraded Limit",
"type": "integer"
},
"cursor": {
"anyOf": [
{
"maxLength": 500,
"type": "string"
},
{
"type": "null"
}
],
"default": null,
"description": "Opaque cursor returned by a previous call; do not construct it.",
"title": "Cursor"
}
},
"title": "McpTargetsPreviewInputV2",
"type": "object"
}