Tools reference

Every Aurora MCP tool is invoked via the standard Model Context Protocol tools/call JSON-RPC method. Your assistant negotiates the MCP handshake for you; this page documents the argument shapes and response structure for reference.

searchMatters

Search engagements accessible to the signed-in Consilio user.

Arguments

Name	Type	Required	Notes
query	string	yes	Free text (client name, descriptor) or an alpha-code.
status	'active' | 'closed' | 'all'	no	Defaults to active.

Alpha-code format: one letter followed by 5 or 6 digits — e.g. A12345, B204891. The tool validates this format before any upstream call.

Response (abbreviated)

[
  {
    "alpha_code": "A12345",
    "name": "Blackstone SEC Investigation",
    "status": "active",
    "client": "Blackstone Inc.",
    "summary": "..."
  }
]

searchWorkspaces

List every workspace inside a matter.

Arguments

Name	Type	Required	Notes
alpha_code	string	yes	Matter alpha-code (e.g. A12345).

Response

[
  { "id": "ws_01HZ...",  "name": "Privileged review", "doc_count": 4123 }
]

searchDocuments

Invoke AI Investigate — Consilio’s retrieval-augmented, agentic investigation engine — inside a matter’s workspace(s). The query is a natural-language question, not a keyword search; AI Investigate refines and iterates on the lookup until it has enough material to answer well.

Arguments

Name	Type	Required	Notes
alpha_code	string	yes	Matter alpha-code.
query	string	yes	Natural-language question or topic to investigate.
workspace_id	string	no	Required only if the matter has more than one workspace. Omit to let the tool auto-scope.

When workspace_id is omitted and the matter has multiple workspaces, the tool returns a disambiguation list instead of results — your assistant can then call again with the chosen workspace.

Response (investigation results)

AI Investigate returns a natural-language answer alongside the supporting evidence it surfaced. Each citation traces back to a specific document and passage:

{
  "answer": "The settlement was proposed on 2026-02-14, citing material weakness in...",
  "citations": [
    {
      "doc_id": "doc_01HZ...",
      "title": "Settlement letter — 2026-02-14",
      "snippet": "...as discussed, we propose <em>settlement</em> terms...",
      "reference": "A12345 · Privileged review · doc_01HZ..."
    }
  ]
}

Common response envelope

Every tool returns a JSON array (or object) inside the standard MCP content wrapper:

{
  "content": [
    { "type": "text", "text": "[{...}, {...}]" }
  ]
}

Your assistant parses the JSON text payload and uses it to answer your prompt.