Atominvest

Returns investment transaction values and financial metric values for a fund as of a specific date. One result object is returned per type requested, inside response.output[]. BEFORE CALLING: - Get investmentId from list_investments. Never guess or construct an ID. - investmentId is a portfolioId, NOT the Investment ID from list_investments. To resolve a fund name: call list_investments → get the Investment ID → call get_investment with that ID → read portfolioId from the result → pass that as investmentId. - Get assetId from list_assets if narrowing to a specific asset. HOW TO PASS TYPES: - Pass as a comma-separated string e.g. "SUBSCRIPTION,NAV,DISTRIBUTION". - Results come back in the same order as types were passed. - Available types: SUBSCRIPTION, DISTRIBUTION, NAV, RETURN_OF_COST, ACCRUED_INCOME, COMMITMENT, EXPENSES, GROSSIRR, GROSS_MULTIPLE, REALIZED_MULTIPLE, REALIZED_CAPITAL, REALIZED_INCOME, VALUATION. WHAT EACH TYPE MEANS: - SUBSCRIPTION: capital called/invested - DISTRIBUTION: cash returned to investors - NAV: net asset value - GROSS_MULTIPLE: (NAV + distributions) / subscriptions - GROSSIRR: gross internal rate of return - RETURN_OF_COST: capital returned against cost basis - ACCRUED_INCOME: income accrued but not yet distributed - COMMITMENT: total committed capital - EXPENSES: fund expenses - REALIZED_MULTIPLE: distributions / subscriptions - REALIZED_CAPITAL: realised capital - REALIZED_INCOME: realised income - VALUATION: valuation transaction amount CURRENCY — IMPORTANT DEFAULT: - Default is "security" if currency is NOT passed. - To get fund reporting currency values, explicitly pass currency="fund". CUMULATIVE vs PERIOD: - isCumulative=true: cumulative value up until the date specified. - isCumulative=false (default): value for the period ending on the date only. - The following types ignore isCumulative — passing it has no effect: GROSSIRR, GROSS_MULTIPLE, NAV, REALIZED_MULTIPLE. - VALUATION rejects isCumulative=true (the call errors) — a valuation is a point-in-time snapshot, not a flow, so summing successive valuations has no financial meaning. Omit isCumulative or pass "false" instead. Request VALUATION in a separate call if you also need cumulative values for other types. - For SUBSCRIPTION, COMMITMENT, EXPENSES: Always pass isCumulative=true unless the user explicitly asks for a specific period's activity only. READING THE RESPONSE: - Results are in response.output — NOT at the root level. - response.filters echoes back what was requested (metadata only). - response.rateLimits is also metadata — ignore when reporting results. - Only investment, date, type, and value are always present in each output object. Optional fields (asset, security, currency, isCumulative, isInverted) only appear in the output if they were included in the request. ZERO VALUE WARNING: - A value of 0 does NOT confirm the metric is zero — it may mean no transactions of that type exist for the fund/date combination. - Do not report 0 as a definitive figure. Qualify it as "no data found or zero." SCOPE: - Without assetId: values cover the entire fund across all assets. - With assetId: values are filtered to that specific asset within the fund.

get_investment_transactions

Get the full definition of a metric by its ID, including the formula field for formula metrics. The 'category', 'subCategory', 'tags' and 'libraries' fields are returned as human-readable names, not IDs. 'tags' and 'libraries' are arrays of names. If a name could not be resolved the raw ID is returned instead — treat any 24-character hex value in these fields as unresolved and do not present it to the user as a name. RESOLVED REFERENCE FIELDS — DO NOT PASS THESE TO OTHER TOOLS: - 'assetId' holds the asset NAME, not an ID. Do NOT pass it to get_metric_data, which requires a real asset UUID — get that from list_assets instead. - 'parentMetricId' holds the parent metric's NAME, not an ID. Do NOT pass it back into get_metric. To get the parent's ID, look its name up in list_metrics. (list_metrics returns this same field as a raw ObjectId — the two tools differ.) Use this when the user asks to break down or analyse a formula metric: 1. Call this tool with the formula metric's ID. 2. Read the returned 'formula' field to identify component metric IDs. 3. Call get_metric_data for each component metric ID separately. Do NOT infer formula components from domain knowledge — only use what this endpoint returns. The metricId passed here must come from an EXACT (case-insensitive) name match found via list_metrics — do not guess an ID for a similarly-named metric.

get_metric

Get time-series metric data for a specific asset and metric. IMPORTANT WORKFLOW: 1. Call list_assets to find the asset ID. 2. Call get_asset_details to retrieve the asset's yearEnd value (asset.metricSettings.yearEnd). 3. Call list_metrics to find the metric ID and period. 4. Call this tool with those values. TABLE INDEX MODES: - Use tableIndex=actuals (default) for actual reported data. Do NOT send scenario. - Use tableIndex=scenarios for scenario/forecast data. MUST send scenario (UUID from list_scenarios). QUERY MODES — choose exactly ONE: A) Point-in-time: provide asOfDate only. B) Date range: provide both filterDateFrom AND filterDateTo. Do NOT mix query modes. METRIC NAME MATCHING: - The metricId passed here must come from an EXACT (case-insensitive) match between the user's requested metric name and a 'name' field returned by list_metrics. - If the user's requested metric name has no exact match in list_metrics, do NOT substitute a similarly-named metric (e.g. do not return "Net Profit" data when the user asked for "Net Income"). - Instead, state explicitly that no exact match was found, list the closest candidate metric name(s) that do exist for that asset, and ask the user to confirm which one they meant before calling this tool.

get_metric_data

Get full details for a specific person by ID.

get_person

Returns accounting transaction ledger records — the same accounting transactions as list_accounting_transactions, but split out per investor. If a $100,000 distribution is allocated across 5 LPs, that's 1 transaction and 5 ledger rows, one per investor's share. BEFORE CALLING: - fundId is required. Get it from list_investments. FILTERING: - fundId (required) — filters to a single fund. - updatedAfter — ISO date, only ledger rows updated after this timestamp. - dateFrom / dateTo — transaction date range. Either or both. ISO date (YYYY-MM-DD accepted). READING THE RESPONSE: - Results are in response.output — NOT at the root level. - allocation and transactionType come back as raw ids only when resolveIds=false. - resolveIds (default true) resolves fundId, entity, asset, security, allocation, transactionType, and transactionTags from raw ids to readable names in place. parentTransaction stays an id — use it to match rows from list_accounting_transactions. - The transaction type name is in the transactionType column. PAGINATION: - Supported params: limit (max records per page), offset (records to skip), page (page number). - Use limit + offset, or limit + page — not both offset and page together. - Without limit, all records are returned in one response. Max limit is 1000. The response has no pagination/hasMore object and no total count: keep paging until a page returns fewer rows than limit.

list_accounting_ledgers

Returns fund-level accounting transaction records — capital calls, distributions, management fees, expenses. One entry per transaction event. For the per-investor breakdown of these same transactions, use list_accounting_ledgers instead. BEFORE CALLING: - fundId is required. Get it from list_investments. FILTERING: - fundId (required) — filters to a single fund. - updatedAfter — ISO date, only transactions updated after this timestamp. - dateFrom / dateTo — transaction date range. Either or both. ISO date (YYYY-MM-DD accepted). READING THE RESPONSE: - Results are in response.output — NOT at the root level. - allocation and transactionType are already full populated objects by default, regardless of resolveIds. - resolveIds (default true) additionally resolves fundId, asset, security, and transactionTags from raw ids to readable names in place. parentValuationTransaction stays an id — use it to match rows from list_asset_transactions. Set resolveIds=false to get the raw ids instead. - The transaction type name is in the transactionType.transactionType column. PAGINATION: - Supported params: limit (max records per page), offset (records to skip), page (page number). - Use limit + offset, or limit + page — not both offset and page together. - Without limit, all records are returned in one response. Max limit is 1000. The response has no pagination/hasMore object and no total count: keep paging until a page returns fewer rows than limit.

list_accounting_transactions

Find an investor account ID from a human-readable account name. Use this ONLY for ID lookup — do NOT use it to answer which entities or people belong to an account. After finding the ID, call get_account for full details. The account name column is "investorName" — accounts have no "name" field. Returns name/ID columns by default — call get_account for full record details (it accepts comma-separated IDs, max 300), or set allFields only when the request genuinely requires additional fields for many accounts at once. PAGINATION: - Use limit (rows per page, max 1000) and page. An unfiltered call defaults to the first 500 rows. - The response includes a "pagination" object: { limit, page, returned, hasMore }. - Follow pagination.hasMore to decide whether to request the next page; equivalently, returned == limit means there is likely another page. - Do NOT combine filterString with limit/page. Filtered results are capped server-side by this tool and report a totalMatches count instead — refine filterString rather than paging.

list_accounts

List all assets (portfolio companies / investments). Supports filtering by name. Use this to find an asset ID before querying its metrics or details. Always returns table format.

list_assets

The authoritative source for onboarding/KYC progress on commitments. Use this for: - Which investors onboarding into a fund or entity haven't finished KYC - What onboarding stage a commitment is stuck on, and since when - Which subscription documents are still pending approval - Overall onboarding status per commitment Do NOT use list_commitments for onboarding questions — it does not return stage or status data. Fund, entity, account, and stage names are already resolved and included by default — no separate lookup needed. PAGINATION: - Use limit (rows per page, max 1000) and page. - pagination.hasMore / pagination.totalPages tells you whether to request another page.

list_commitment_onboarding

The AUTHORITATIVE source for commitment analytics. Use this for: - Exact commitment amounts per entity or account - Commitment statuses (committed / pending) - Commitment dates - Total invested amounts by entity, account, or fund - Mapping commitments to entities and funds Do NOT use embedded commitment arrays from account or entity records for any of the above — always use this endpoint instead. WORKFLOW NOTES: - To get commitments for an account: pass account=<accountId> - To get commitments for a single entity: pass entity=<entityId> - Resolve entity IDs to names using list_entities - Resolve fund IDs to names using list_investments - Flag blanks, tiny values, duplicates, and outliers rather than silently normalising them

list_commitments

List companies by name, to resolve a company name into a company ID. Returns name/ID columns by default — call get_company for full record details (it accepts comma-separated IDs, max 300), or set allFields only when the request genuinely requires additional fields for many companies at once. PAGINATION: - Use limit (rows per page, max 1000) and page. An unfiltered call defaults to the first 500 rows. - The response includes a "pagination" object: { limit, page, returned, hasMore }. - Follow pagination.hasMore to decide whether to request the next page; equivalently, returned == limit means there is likely another page. - Do NOT combine filterString with limit/page. Filtered results are capped server-side by this tool and report a totalMatches count instead — refine filterString rather than paging.

list_companies

List entities by name, to resolve an entity name into an entity ID. Returns name/ID columns by default. For entity attributes (type, domicile, source of wealth, registration, legal structure, risk fields, etc.) call get_entity with the ID — that is the authoritative source. Set allFields here only when the request genuinely requires attributes for many entities at once; for a single entity, always prefer get_entity. Do NOT infer entity attributes from account-embedded entity objects — resolve the entity ID and call get_entity instead. PAGINATION: - Use limit (rows per page, max 1000) and page. An unfiltered call defaults to the first 500 rows. - The response includes a "pagination" object: { limit, page, returned, hasMore }. - Follow pagination.hasMore to decide whether to request the next page; equivalently, returned == limit means there is likely another page. - Do NOT combine filterString with limit/page. Filtered results are capped server-side by this tool and report a totalMatches count instead — refine filterString rather than paging.

list_entities

List all available fund-level metrics (KPIs like NAV, IRR, DPI, and custom fund KPIs) for a specific fund. Returns enriched data with metric IDs and names. Use this to discover which fund metrics are available before querying their data. This list returns identifying fields only: '_id', 'name', 'dataType', 'isParentMetric', 'parentMetricId', 'carryType', 'createdAt' and 'updatedAt'. The 'category', 'subCategory', 'tags' and 'libraries' fields are NOT included here — call get_metric for a specific metric if you need them. Do not group, filter or report on those fields based on this tool's output. METRIC NAME MATCHING: When resolving a user-requested metric name to a metric ID, require an EXACT (case-insensitive) match against the returned 'name' field. Do NOT treat a similarly-named metric as a match (e.g. "Net Income" is NOT a match for "Net Profit"). If no exact match exists, do not guess — report that no exact match was found and list the closest candidate names instead.

list_fund_metrics

List investments (funds). Use this mainly to resolve a fund ID to a human-readable fund name. Do not send allFields or resolveIds — only outputFormat and optional filterString.

list_investments

List all available metrics (KPIs like Revenue, EBITDA, Headcount, etc.). Can be filtered by a specific asset. Returns enriched data with metric IDs and names. Use this to discover which metrics are available before querying their data. This list returns identifying fields only: '_id', 'name', 'dataType', 'isParentMetric', 'parentMetricId', 'carryType', 'createdAt' and 'updatedAt'. The 'category', 'subCategory', 'tags' and 'libraries' fields are NOT included here — call get_metric for a specific metric if you need them. Do not group, filter or report on those fields based on this tool's output. 'parentMetricId' is a raw ObjectId in this list (note: get_metric returns the same field as a NAME instead). METRIC NAME MATCHING: When resolving a user-requested metric name to a metric ID, require an EXACT (case-insensitive) match against the returned 'name' field. Do NOT treat a similarly-named metric as a match (e.g. "Net Income" is NOT a match for "Net Profit"). If no exact match exists, do not guess — report that no exact match was found and list the closest candidate names instead.

list_metrics

List persons (contacts/individuals) by name or filter. Use this when enrichment beyond account-linked contacts is needed. Always returns allFields, table format.

list_persons

List available scenarios (forecasts/projections) for an asset or fund. Use this to find a scenario UUID before calling get_metric_data with tableIndex=scenarios.

list_scenarios

The source for Smart Workflow progress — general stage-based workflows attached to accounts, entities, commitments, assets, or people. This is separate from fund onboarding. For fund onboarding workflows, use list_commitment_onboarding instead. Use this tool to: - Check what stage an account, entity, commitment, asset, or person is at in a workflow - See what is left to complete or review for a specific workflow assignment - Check the overall status of each workflow entry Use list_smart_workflow_templates first to find the template ID by name. Template and relationship names are resolved by default. Stage names are returned in stages[]; keys inside stagesData remain raw stage IDs. Per-stage status and overallStatus are included by default. Set includeStageStatuses=false for a leaner response when you only need relationship or workflow-existence data. Status values (stages[].status, overallStatus, and stagesData[stageId].status): - resolveIds=true (default): Not Started, In Progress, Submitted/Under Review, Completed, Not required, Archived - resolveIds=false: not_started, in_progress, submitted, approved, not_required, archived (stagesData may also contain the legacy value completed, equivalent to approved) Known limitations: - For account-linked workflows, the entity behind the account is not resolved; for asset-linked workflows, the account/entity behind the asset is not resolved. This is a platform limitation, not specific to this tool. - For commitment-linked workflows, the account and entity names are returned as accountName / entityName inside relationshipFields. - For kyc_document_upload stages, raw document data in stagesData uses KCYDocuments / KCYDocumentsSettingId (a known typo in the underlying schema, not KYCDocuments). Pagination: - Use limit (rows per page, max 1000) and page. - pagination.hasMore / pagination.totalPages tells you whether to request another page.

list_smart_workflow_entries

List Smart Workflow templates (the checklist definitions) by name, to resolve a template name into a templateId. Use list_smart_workflow_entries to see actual progress against a template. PAGINATION: - Use limit (rows per page, max 1000) and page. - pagination.hasMore tells you whether to request another page; pagination.totalPages is only present when page was passed.

list_smart_workflow_templates

Search across multiple object types by name in a single call. Returns a flat list of matches, each tagged with its objectType (asset, entity, person, company, account, investment) and id. USE THIS FIRST when the user refers to something by name but you do not yet know its type or ID — e.g. "how is Algolia doing?" or "show me everything for Altman". It resolves a fuzzy name into concrete { id, name, objectType } records so you know which get_* tool to call next. This tool ONLY finds records and their IDs/types — it does NOT return details. After identifying the right record, call the matching tool (get_asset_details, get_entity, get_person, get_company, get_account, get_investment) with the returned id. Matching is case-insensitive partial match on the object name (accounts match on investor name).

search

Returns raw individual asset transaction records for a fund, asset, or security. DIFFERENCE FROM get_investment_transactions: - get_investment_transactions returns aggregated metric values (NAV, IRR, subscription totals). - list_asset_transactions returns the raw individual transaction records themselves. Use this tool when you need the actual transaction entries, not rolled-up metrics. BEFORE CALLING: - investment is a Portfolio ID, NOT the Fund Vehicle ID returned by list_investments. To filter by a fund name: call list_investments → get the Investment ID → call get_investment with that ID → read portfolioId from the result → pass that as investment. - Get asset ID from list_assets if filtering by asset. - All params are optional — omitting all returns every transaction in the environment. FILTERING: - Pass investment, asset, security, or type to narrow results. - Pass updatedAfter (ISO date) to get only transactions updated after that point — matches updatedAt, not createdAt, so edits to existing records are captured. - IMPORTANT: the fund filter param is called "investment", not "investmentId". - type must be one or multiple of the enum values — the backend validates and throws if an invalid value is passed. Custom transaction type names configured in the admin UI are also valid — do not flag unrecognised type values as invalid or non-standard. READING THE RESPONSE: - Results are in response.output — NOT at the root level. - response.filters is {} when no filter params were passed — this is normal. - response.rateLimits is metadata — ignore when reporting results. - Each record has two amount fields: - amount: transaction amount in the security's native currency. - amountInfundCurrency: transaction amount converted to fund reporting currency. - amount can be negative — this is normal and represents returns or redemptions. - units and unitPrice are only populated for transactions tied to a specific security (e.g. SUBSCRIPTION, VALUATION) — they are empty for types with no unit-level basis, such as EXPENSES and ACCRUED_INCOME. - unitPrice has no fund-currency equivalent (unlike amount → amountInfundCurrency) — it is always expressed in the security's native currency. - reportingDate is in YYYY-MM-DD format — not a full ISO timestamp. - Results are ordered by createdAt, not reportingDate. Do not assume chronological order. MATCHING FIELDS — IMPORTANT: - The "investment" filter param matches against the "portfolio" field in the response, NOT the "fund" field. - "portfolio": the Portfolio ID — matches the portfolioId from get_investment, NOT the ID from list_investments. - "fund": the parent Fund Vehicle ID — one level above, a different entity. - Do not cross-reference the passed investment ID against "fund" in results — they will not match. ZERO AMOUNT WARNING: - amount: 0 does NOT mean the transaction is zero in all currencies. A record can have amount: 0 with amountInfundCurrency: 30,000,000 when the security currency differs from the fund currency. - Always check amountInfundCurrency for the fund-level figure. DYNAMIC FIELDS: - dynamicFields[] contains custom fields configured per client environment. - Use renderedValue (not outputValue) when displaying dynamic field values to users. outputValue can be a raw ID array — renderedValue is always human-readable. - dynamicFields can be an empty array [] — this is normal. - The "type" field inside each dynamicField object is the UI field type (e.g. "selectMultiple") — not the transaction type. DELETED TRANSACTIONS: - By default, deleted transactions are excluded from results. - Pass includeDeleted=true to include them. Deleted records will have isDeleted: true. - Deletion tracking is only available on environments upgraded after 22 May 2026. Transactions deleted before that date cannot be retrieved even with includeDeleted=true. PAGINATION: - Supported params: limit (max records per page), offset (records to skip), page (page number). - Use limit + offset, or limit + page — not both offset and page together. - The response includes a "pagination" object: { limit, offset, hasMore }. - Use pagination.hasMore to know when to stop — if true, there are more records to fetch. - response.count is the number of records in the current page, not the total record count. - Without limit, the endpoint appears to return all records in one response.