Argus Data Store MCP Tools#
These tools let the LLM retrieve Argus datastore metadata and centrally managed datastore entries used by detection rules.
They support two common tasks:
fetch a datastore descriptor by datastore name
fetch one datastore entry by datastore name, entry key, and customer scope
General Notes#
Use these tools when you want the LLM to inspect an Argus datastore descriptor or look up a specific datastore entry value.
Datastores are used by detection rules and may be structured as either maps or lists.
The descriptor controls whether entry lookup is valid and which customer value must be used.
The LLM should fetch the datastore descriptor before trying to fetch entries if the descriptor has not already been checked.
Only datastores with
behaviourType = CENTRALcan have entries fetched through these MCP tools.If
behaviourType = LOCAL, the datastore descriptor exists, but entries are managed inside distributed detection pipeline environments and are not accessible through MCP tools.If
globalData = true, entry lookup must usecustomerID = 0because the datastore is global.If
globalData = false, entry lookup must use a valid Argus customer ID because the datastore is customer-scoped.Tool results are live Argus API results. The LLM should treat them as the source of truth and not infer missing datastore values.
Detection Rule Context#
Detection rules can refer to datastores in patterns such as:
global map lookup by datastore name
customer map lookup by datastore name and customer ID
global list lookup by datastore name
customer list lookup by datastore name and customer ID
global or customer-scoped entry lookup by datastore name and entry key
The MCP tools expose descriptor lookup and individual entry lookup. They do not expose full datastore enumeration in this tool surface.
Descriptor Fields That Drive Entry Lookup#
Descriptor field |
Allowed values |
How it affects later lookup |
|---|---|---|
|
|
|
|
|
Describes whether the datastore stores key-value data or a collection of values. Entry lookup still uses an entry key. |
|
|
|
Tool: getArgusDataStoreDescriptorByName#
Purpose
Fetch the Argus datastore descriptor for a datastore name. The descriptor explains the datastore’s structure, scope, and management model.
Arguments
Field |
Description |
Type |
Default |
Required |
|---|---|---|---|---|
|
Name of the datastore descriptor to fetch. |
|
None |
Yes |
Usage Notes
Blank or missing datastore names fail validation.
The descriptor should be checked before entry lookup so the LLM does not assign the wrong customer scope or try to fetch a
LOCALdatastore entry.For
LOCALdatastores, the LLM should explain that the descriptor exists but entries cannot be fetched through MCP.For
CENTRALdatastores, the LLM can continue to entry lookup when it also has the entry key and the correct customer scope.
Tool: getArgusDataStoreEntry#
Purpose
Fetch one Argus datastore entry by datastore name, entry key, and customer scope. If the entry is not found, the returned data may be null.
Arguments
Field |
Description |
Type |
Default |
Required |
|---|---|---|---|---|
|
Name of the datastore. Must match the descriptor name. |
|
None |
Yes |
|
Key of the datastore entry to fetch. |
|
None |
Yes |
|
Customer scope for the entry lookup. Use |
|
None |
Yes |
Usage Notes
Blank or missing datastore names fail validation.
Blank or missing entry keys fail validation.
The LLM should verify the datastore descriptor first if it has not already done so.
Entry lookup is only appropriate for
CENTRALdatastores.Do not use
customerID = 0for customer-scoped datastores whereglobalData = false.Do not use a real customer ID for global datastores where
globalData = true; use0instead.If the entry is not found, the tool can return a response where the entry data is
null; this should be reported plainly instead of guessing a value.