Blumira API Patterns

Blumira API Patterns

Overview

Blumira exposes a REST API at https://api.blumira.com/public-api/v1 with two path groups: /org/* for direct organization access and /msp/* for MSP multi-tenant operations. The MCP server wraps these into tool calls, but understanding the underlying patterns helps construct effective queries.

Key Concepts

Authentication

Blumira uses JWT tokens for authentication. The token is passed via the X-Blumira-JWT-Token header (MCP Gateway) or as a Bearer token directly against the API.

Authorization: Bearer <JWT_TOKEN>

Alternatively, for Pax8 integrations:

pax8ApiTokenV1: <PAX8_TOKEN>

Important: JWT tokens have expiration times. If you receive 401 errors, the token may need to be regenerated from the Blumira portal.

Dual Path Groups

Path Group	Prefix	Use Case
Organization	/org/*	Direct access to a single organization's data
MSP	/msp/*	Multi-tenant access across managed accounts

Organization tools (blumira_findings_*, blumira_agents_*, blumira_users_*) operate on the authenticated org. MSP tools (blumira_msp_*) require MSP-level credentials and can target specific accounts.

Rich Filtering Syntax

Blumira supports powerful query filters appended to field names:

Operator	Suffix	Example	Description
Equals	.eq	status.eq=10	Exact match
In	.in	severity.in=HIGH,CRITICAL	Match any in list
Greater than	.gt	created.gt=2025-01-01	Greater than
Less than	.lt	created.lt=2025-12-31	Less than
Contains	.contains	name.contains=ransomware	Substring match
Regex	.regex	name.regex=^brute	Regex match
Negation	! prefix	!status.eq=30	Negate any filter

Combining filters: Multiple filters are ANDed together:

GET /org/findings?status.eq=10&severity.in=HIGH,CRITICAL&created.gt=2025-01-01

Pagination

All list endpoints support pagination parameters:

Parameter	Description	Default
page	Page number (1-indexed)	1
page_size	Results per page	25
limit	Max total results	—
order_by	Sort field (prefix - for descending)	varies

Responses include pagination metadata:

{
  "data": [...],
  "links": {
    "next": "/org/findings?page=2&page_size=25",
    "prev": null
  },
  "meta": {
    "total": 142,
    "page": 1,
    "page_size": 25
  }
}

MCP Navigation Tools

The MCP server includes stateful navigation tools:

• blumira_navigate — Navigate to a specific resource or view
• blumira_status — Show current navigation context
• blumira_back — Return to previous context

These help maintain context when drilling into findings, devices, or accounts.

Common Workflows

Filtered Finding Query

1. Use blumira_findings_list with filter parameters
2. Narrow results with status, severity, and date filters
3. Page through results if needed
4. Drill into specific findings with blumira_findings_get

MSP Cross-Account Query

1. Use blumira_msp_accounts_list to enumerate accounts
2. Use blumira_msp_findings_all for cross-account finding overview
3. Filter to specific account with blumira_msp_findings_list
4. Drill into per-account details as needed

Error Handling

401 Unauthorized

Cause: JWT token is expired, invalid, or missing Solution: Regenerate the token from Blumira Portal > Settings > API Access. Verify BLUMIRA_JWT_TOKEN is set.

403 Forbidden

Cause: Token lacks permissions for the requested resource (e.g., org token used for MSP endpoints) Solution: Ensure the token has appropriate scope. MSP endpoints require MSP-level credentials.

404 Not Found

Cause: Resource ID doesn't exist or is not accessible from the current scope Solution: Verify the ID and ensure the token has access to the target organization.

429 Rate Limited

Cause: Too many requests in a short period Solution: Back off and retry after a delay. Use pagination to reduce request volume.

422 Validation Error

Cause: Invalid filter syntax or parameter values Solution: Check filter operator syntax (.eq, .in, etc.) and ensure values match expected types.

Best Practices

• Use pagination (page_size=50) for large datasets instead of fetching everything at once
• Combine filters to narrow results before fetching — don't over-fetch and filter client-side
• Use order_by=-created to get most recent items first
• Cache account lists when doing MSP operations to avoid repeated lookups
• Always handle pagination — check meta.total to know if more pages exist

Related Skills

• Findings — Finding lifecycle management
• MSP — MSP multi-tenant operations
• Agents — Device and agent management