Detect privileged content

Analyzes text or vault documents for legal privilege review. Detects attorney-client privilege, work product doctrine, and common interest privilege.

Returns structured review flags with confidence scores and policy-friendly rationale suitable for discovery workflows and privilege logs. This endpoint is an AI-assisted triage tool and does not replace attorney judgment.

Size Limit: Maximum 200,000 characters (larger documents rejected).

Permissions: Requires chat permission. When using document_id, also requires vault permission.

Note: When analyzing vault documents, results are automatically stored in the document’s privilege_analysis metadata field.

POST

/

privilege

/

v1

/

detect

{
  "privileged": true,
  "confidence": 0.5,
  "categories": [
    {
      "type": "<string>",
      "detected": true,
      "confidence": 123,
      "rationale": "<string>",
      "indicators": [
        "<string>"
      ]
    }
  ],
  "policy_rationale": "<string>"
}

Authorizations

Authorization

string

header

required

API key starting with sk_case_

Body

application/json

content

string

Text content to analyze (required if document_id not provided)

document_id

string

Vault object ID to analyze (required if content not provided)

vault_id

string

Vault ID (required when using document_id)

categories

enum<string>[]

Privilege categories to check. Defaults to all: attorney_client, work_product, common_interest

Available options:

attorney_client,

work_product,

common_interest

jurisdiction

enum<string>

default:US-Federal

Jurisdiction for privilege rules

Available options:

US-Federal

include_rationale

boolean

default:true

Include detailed rationale for each category

model

string

default:casemark/core-large

LLM model to use for analysis

Response

Privilege analysis completed

privileged

boolean

required

Whether any privilege was detected

confidence

number

required

Overall confidence score (0-1)

Required range: 0 <= x <= 1

categories

object[]

required

Show child attributes

recommendation

enum<string>

required

Recommended action for discovery

Available options:

withhold,

redact,

produce,

review

policy_rationale

string

required

Policy-friendly explanation for privilege log

Generate AI-powered answers from web searchGenerate comprehensive answers to questions using web search results. Supports two modes: native provider answers or custom LLM-powered answers using Case.dev's AI gateway. Perfect for legal research, fact-checking, and gathering supporting evidence for cases.

⌘I

{
  "privileged": true,
  "confidence": 0.5,
  "categories": [
    {
      "type": "<string>",
      "detected": true,
      "confidence": 123,
      "rationale": "<string>",
      "indicators": [
        "<string>"
      ]
    }
  ],
  "policy_rationale": "<string>"
}