About Blumira
Blumira is an MCP server that gives AI assistants structured access to Blumira SIEM data and operations across findings, devices, users, resolutions, and MSP accounts. It supports rich filtering, multi-tenant access, and domain-based tool discovery.
Key features
- 01Query and manage security findings
- 02View devices, agent keys, and users
- 03Access MSP account data
- 04Filter with multiple field operators
- 05Support stdio and HTTP transports
Use cases
- 01Filter and investigate security findings
- 02Assign finding owners or add comments
- 03Inspect devices and agent keys
- 04Review findings and users across MSP accounts
MCP support
The MCP server exposes Blumira API data and operations as tools and resources for MCP clients, with local stdio and remote HTTP transports.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
docker run -i --rm -e BLUMIRA_JWT_TOKEN -e MCP_TRANSPORT -e AUTH_MODE -e LOG_LEVEL ghcr.io/wyre-ai/blumira-mcp:v1.2.16BLUMIRA_JWT_TOKENSecretBlumira API JWT token used to authenticate requestsMCP_TRANSPORTTransport mode for the server. Set to 'stdio' for local CLI use; the image defaults to 'http' for gateway hosting.AUTH_MODESecretCredential source: 'env' reads vars locally, 'gateway' expects header injection from the WYRE MCP Gateway.LOG_LEVELLog verbosity: debug, info, warn, errorLocal servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
ghcr.io/wyre-ai/blumira-mcp:v1.2.16stdioTools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEBlumira MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with structured access to Blumira SIEM platform data and operations.
Note: This project is maintained by Wyre Technology.
Quick Start
Claude Desktop — download, open, done:
- Download
blumira-mcp.mcpbfrom the latest release - Open the file (double-click or drag into Claude Desktop)
- Enter your Blumira JWT token when prompted
No terminal, no JSON editing, no Node.js install required.
Claude Code (CLI):
claude mcp add blumira-mcp \
-e BLUMIRA_JWT_TOKEN=your-jwt-token \
-- npx -y github:WYRE-AI/blumira-mcp
See Installation for Docker and from-source methods.
Features
- 🔌 MCP Protocol Compliance: Full support for MCP resources and tools
- 🛡️ Comprehensive SIEM Coverage: Tools spanning findings, agents/devices, users, resolutions, and MSP account management
- 🔍 Decision-Tree Navigation: Start with
blumira_navigateto explore domains, then dynamically load domain-specific tools - 🏢 MSP Multi-Tenant Support: Full MSP endpoint coverage for managing findings, agents, and users across accounts
- 🔒 Secure Authentication: JWT token or API key (
pax8ApiTokenV1) authentication - 🌐 Dual Transport: Supports both stdio (local) and HTTP Streamable (remote/Docker) transports
- 📦 MCPB Packaging: One-click installation via MCP Bundle for desktop clients
- 🐳 Docker Ready: Containerized deployment with HTTP transport and health checks
- ⚡ Rate Limiting: Built-in rate limiter respects Blumira API limits
- 🔎 Rich Filtering: Support for
.eq,.in,.gt,.lt,.contains,.regex, and negation operators
Installation
Option 1: MCPB Bundle (Claude Desktop)
The simplest method — no terminal, no JSON editing, no Node.js install required.
- Download
blumira-mcp.mcpbfrom the latest release - Open the file (double-click or drag into Claude Desktop)
- Enter your Blumira JWT token when prompted
For Claude Code (CLI), one command:
claude mcp add blumira-mcp \
-e BLUMIRA_JWT_TOKEN=your-jwt-token \
-- npx -y github:WYRE-AI/blumira-mcp
Option 2: Docker
docker compose up
Compose will not start until CONDUIT_S2S_SECRET is set. See Docker Deployment.
Or pull the pre-built image:
# Generate the secret first: openssl rand -hex 32
docker run -d \
-e CONDUIT_S2S_SECRET \
-e BLUMIRA_JWT_TOKEN \
-e AUTH_MODE=env \
-p 127.0.0.1:8080:8080 \
ghcr.io/wyre-ai/blumira-mcp:latest
The image defaults to AUTH_MODE=gateway and binds 0.0.0.0 inside the container. HTTP startup refuses to run unless CONDUIT_S2S_SECRET is a real secret (empty values and known placeholders such as replace-with-a-real-secret exit non-zero). Publish the port on 127.0.0.1 (as above). In gateway mode, send vendor credentials on each /mcp request (X-Blumira-Client-ID + X-Blumira-Client-Secret, or X-Blumira-JWT-Token); the server does not fall back to environment credentials. Set AUTH_MODE=env only for a single-tenant container that should use BLUMIRA_JWT_TOKEN.
Every /mcp request must also send X-Gateway-S2S. The check is verifyS2sHeader in src/s2s-verify.ts: the header value is t=<unix seconds>,v1=<hex>, where the hex is HMAC-SHA256 of the literal string t=<unix seconds> keyed with CONDUIT_S2S_SECRET (64 lowercase hex characters). The timestamp must be within 300 seconds of the server clock. A missing or invalid header is HTTP 401, before vendor credentials are read.
TS=$(date +%s)
SIG=$(printf 't=%s' "$TS" | openssl dgst -sha256 -hmac "$CONDUIT_S2S_SECRET" -hex | awk '{print $NF}')
curl -sS http://127.0.0.1:8080/mcp \
-H 'Content-Type: application/json' \
-H 'Accept: application/json, text/event-stream' \
-H "X-Gateway-S2S: t=${TS},v1=${SIG}" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'
With AUTH_MODE=env, that signed request uses BLUMIRA_JWT_TOKEN from the container. With AUTH_MODE=gateway, add X-Blumira-JWT-Token or X-Blumira-Client-ID and X-Blumira-Client-Secret as well.
Standalone local use that cannot sign requests: set MCP_ALLOW_INSECURE_DEV=1 and do not set CONDUIT_S2S_SECRET. Startup logs a warning, skips the S2S check, and binds 127.0.0.1 only. The image sets MCP_HTTP_HOST=0.0.0.0, so this bypass also requires -e MCP_HTTP_HOST=127.0.0.1. Any other bind address refuses to start. Do not publish that port off loopback.
Option 3: From Source
git clone https://github.com/WYRE-AI/blumira-mcp.git
cd blumira-mcp
npm ci
npm run build
Configuration
| Variable | Description | Default |
|---|---|---|
BLUMIRA_JWT_TOKEN | JWT token for authentication (stdio and AUTH_MODE=env) | — |
MCP_TRANSPORT | Transport mode (stdio or http) | stdio |
MCP_HTTP_PORT | HTTP server port | 8080 |
MCP_HTTP_HOST | Interface the HTTP server binds | 127.0.0.1 |
CONDUIT_S2S_SECRET | Required for HTTP. HMAC secret checked against X-Gateway-S2S on /mcp. Generate with openssl rand -hex 32. If empty or a known placeholder, the HTTP server logs an error and exits non-zero. The value is never logged. | — |
MCP_ALLOW_INSECURE_DEV | Set to 1 to start HTTP without CONDUIT_S2S_SECRET. Logs a warning, does not enforce S2S, and binds 127.0.0.1 only. A non-loopback MCP_HTTP_HOST refuses to start. Local development only. | unset |
AUTH_MODE | env uses process environment credentials (stdio and single-tenant HTTP). gateway requires per-request vendor headers and never falls back to the environment. The Docker image and Compose file default to gateway. | env when unset; gateway in Docker and Compose |
LOG_LEVEL | Log level (debug, info, warn, error) | info |
/health and /healthz stay unauthenticated and do not read credentials. The stdio transport does not use CONDUIT_S2S_SECRET.
Domains
The server uses decision-tree navigation. Start with blumira_navigate to pick a domain:
| Domain | Tools |
|---|---|
| findings | List findings, get finding, get finding details, resolve finding, assign owners, list/add comments |
| agents | List devices, get device, list agent keys, get agent key |
| users | List users |
| resolutions | List available resolutions |
| msp | List/get accounts, list/get/resolve findings, assign owners, comments, list devices/keys, list users |
Filtering
Blumira supports rich query filtering on list endpoints:
status.eq=10 # Exact match
severity.in=HIGH,CRITICAL # Multiple values
created_at.gt=2026-01-01 # Greater than
name.contains=malware # Substring match
!status.eq=30 # Negation
Pass filters as tool input parameters — the server handles query string construction.
Docker Deployment
Copy .env.example to .env and fill in your credentials, including CONDUIT_S2S_SECRET. Compose will not start if that variable is missing, and it publishes the HTTP port on 127.0.0.1 only.
cp .env.example .env
# Edit .env with your Blumira JWT token and CONDUIT_S2S_SECRET
docker compose up -d
Development
npm ci
npm run build # Build the project
npm run dev # Watch mode
npm run test # Run tests
npm run lint # Type-check
npm run clean # Remove dist/
Contributing
See CONTRIBUTING.md for guidelines.
License
Apache 2.0 — Copyright WYRE Technology
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.