taegis-magic-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@taegis-magic-mcp List high severity alerts from last hour"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
taegis-magic-mcp
An MCP (Model Context Protocol) server that wraps taegis-magic — the Secureworks Taegis XDR CLI and SDK — so that AI assistants and agents can query alerts, events, investigations, threat intelligence, and tenants directly from their context window.
How it works
taegis-magic is the official open-source
CLI and SDK for Secureworks Taegis XDR. It exposes subcommands like
taegis alerts search, taegis investigations search, taegis events search, etc. via
a rich Python SDK backed by the Taegis GraphQL API.
This server wraps those CLI commands behind the MCP tool protocol: an agent calls a tool
(search_alerts, search_events, …), the server runs the appropriate taegis command as
a subprocess, and returns the JSON output to the agent. No Taegis API knowledge is required
on the agent side.
Agent / LLM
│
│ MCP tool call: search_alerts(query="...", region="charlie")
▼
taegis-magic-mcp (this server)
│
│ subprocess: taegis alerts search --cell "..." --region charlie --limit 100
▼
taegis-magic CLI ─────► Secureworks Taegis XDR APIRelated MCP server: wazuh-mcp-server
Requirements
Requirement | Notes |
Python ≥ 3.11 | |
Used to run the server in an isolated environment | |
Must be installed and on |
Authentication
taegis-magic supports two authentication methods:
OAuth client credentials (recommended for automated use)
export TAEGIS_CLIENT_ID=<your-oauth-client-id>
export TAEGIS_CLIENT_SECRET=<your-oauth-client-secret>
export TAEGIS_ENVIRONMENT=<region> # charlie | delta | us1 | us2 | euThe MCP server forwards these as CLIENT_ID / CLIENT_SECRET to the
taegis-sdk-python environment
(the names the SDK reads). TAEGIS_CLIENT_ID / TAEGIS_CLIENT_SECRET are the
namespaced forms used here to avoid collisions with other tools.
Interactive login (recommended for interactive use)
taegis auth loginThe cached token is stored at ~/.taegis_sdk_python/config and reused automatically.
Available MCP tools
Tool | Description |
| Search Taegis alerts using OCSF query syntax |
| Search Taegis security events |
| List open security investigations |
| Search threat intelligence (IOCs, hashes, IPs, domains …) |
| List all tenants visible to the authenticated user |
| Escape hatch — run any |
All tools accept an optional region parameter (overrides TAEGIS_ENVIRONMENT).
Example queries
search_alerts(query="FROM alert EARLIEST=-1d WHERE severity='High'", region="charlie")
search_events(query="FROM event EARLIEST=-4h WHERE source_ip='10.0.0.1'")
search_threat_intel(query="evil.example.com")
run_taegis_command(args="rules search --region charlie --limit 10")Installation
With opencode
Clone or add this repo as a submodule, then add to your opencode.jsonc:
"mcp": {
"taegis": {
"type": "local",
"command": ["uv", "run", "--project", "/path/to/taegis-magic-mcp", "taegis-mcp"],
"environment": {
"TAEGIS_CLIENT_ID": "${TAEGIS_CLIENT_ID}",
"TAEGIS_CLIENT_SECRET": "${TAEGIS_CLIENT_SECRET}",
"TAEGIS_ENVIRONMENT": "${TAEGIS_ENVIRONMENT}"
},
"enabled": false
}
}Set "enabled": true (or toggle at runtime) when you want the Taegis tools in scope.
Standalone
git clone https://github.com/DarkLicornor/taegis-magic-mcp.git
cd taegis-magic-mcp
uv sync
uv run taegis-mcpDevelopment
git clone https://github.com/DarkLicornor/taegis-magic-mcp.git
cd taegis-magic-mcp
uv sync --extra dev
uv run taegis-mcp # run the server
uv run pytest # run tests
uv run ruff check . # lint
uv run mypy src # type-checkSee also
taegis-magic — the underlying Taegis CLI / SDK this server wraps
taegis-sdk-python — the Python SDK taegis-magic is built on
Model Context Protocol — the protocol this server implements
FastMCP — the MCP server framework used
License
MIT
Available Tools
6 toolslist_tenantsA
List all Taegis tenants visible to the authenticated user.
Returns: JSON array of tenant objects.
Inputs: no named parameters. Returns: result
run_taegis_commandA
Run an arbitrary taegis CLI command and return JSON output.
Use this as an escape hatch for commands not yet wrapped as dedicated tools
(e.g. "rules search --region charlie").
Args:
args: Space-separated CLI arguments after the taegis binary,
e.g. "audits search --region charlie --limit 10".
Returns: Raw JSON output from the taegis CLI.
Inputs: args (required). Returns: result
search_alertsA
Search Taegis alerts using OCSF query syntax.
Args:
query: OCSF query string, e.g. "FROM alert EARLIEST=-1d WHERE severity='High'"
region: Taegis environment/region (charlie, delta, us1, us2, eu …).
Defaults to the TAEGIS_ENVIRONMENT env var, then "charlie".
limit: Maximum number of results to return (default 100, max 10 000).
Returns: JSON array of alert objects.
Inputs: query (required), limit, region. Returns: result
search_eventsA
Search Taegis security events.
Args: query: Event query string. region: Taegis environment/region. Defaults to TAEGIS_ENVIRONMENT or "charlie". limit: Maximum number of results (default 100).
Returns: JSON array of event objects.
Inputs: query (required), limit, region. Returns: result
search_investigationsA
List and search open security investigations in Taegis.
Args: region: Taegis environment/region. Defaults to TAEGIS_ENVIRONMENT or "charlie". limit: Maximum number of investigations to return (default 25).
Returns: JSON array of investigation objects.
Inputs: limit, region. Returns: result
search_threat_intelA
Search Taegis threat intelligence (IOCs, hashes, IPs, domains …).
Args: query: Search query — indicator value or free-text search. region: Taegis environment/region. Defaults to TAEGIS_ENVIRONMENT or "charlie".
Returns: JSON array of threat intelligence objects.
Inputs: query (required), region. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Each tool targets a distinct resource type (alerts, events, investigations, tenants, threat intel) with clear search/list semantics. The escape hatch command is explicitly for arbitrary CLI calls, so there's no overlap with the dedicated tools.
All tool names follow a consistent verb_noun snake_case pattern (search_alerts, list_tenants, run_taegis_command). The pattern is uniform across the set with no mixed conventions.
Six tools is well within the ideal 3-15 range. The server covers the core Taegis search use cases without redundancy, and the escape hatch avoids the need for many narrow wrappers.
The dedicated tools cover the most common read/search operations for a security platform. Minor gaps exist (e.g., no dedicated get-by-id tools or investigation management), but the run_taegis_command escape hatch fills any missing functionality, making the surface effectively complete.
Related MCP Connectors
An MCP server that provides an API to LLMs to manage their JumpCloud resources.
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
MCP server for Pentest-Tools.com: run scans, manage findings and reports via your preffered LLM.
Model Context Protocol server for the Apideck Unified API. Connect any MCP-compatible agent framework to 100+ accounting systems, HRIS platforms, file storage providers, and more through one integration. More information https://www.apideck.com/mcp-server
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceUniversal MCP server that wraps any CLI tool, enabling AI assistants to run commands via natural language.1MIT
- AlicenseAqualityBmaintenanceAI-powered MCP server that enables security analysts to query Wazuh SIEM/XDR for alert triage, threat hunting, compliance audits, and incident response through natural language prompts.2813MIT
- FlicenseNot gradedqualityDmaintenanceA robust MCP server that bridges AI development tools with Palo Alto Cortex platforms (XSOAR/XSIAM), enabling natural language-driven development, testing, and automation of security workflows.2-
- AlicenseNot gradedqualityBmaintenanceA modular, multi-transport Model Context Protocol server that connects AI assistants to the CrowdStrike Falcon platform. Query NG-SIEM logs, triage alerts, inspect endpoints, manage detection rules, and audit cloud security posture — all through natural language.14MIT