AnomalyArmor MCP Server for Claude
AnomalyArmor armor-mcp stdio server with 52 consolidated data observability tools for alerts, freshness, schema drift, quality metrics, lineage, and AI recommendations.
Open the source and read safety notes before installing.
Safety notes
- Tools can create alert rules, freshness schedules, metrics, and monitoring configurations.
- manage_asset and related tools may trigger discovery jobs on production sources.
- Alert acknowledgement and resolution changes operational state.
- Review AI recommendations before applying automated monitoring at scale.
Privacy notes
- Table names, schema metadata, alert details, and sample metrics are sent to AnomalyArmor.
- Natural language `ask_question` queries may expose internal data catalog context to the LLM.
- API keys grant access until revoked in AnomalyArmor settings.
Prerequisites
- AnomalyArmor account with an API key from Settings → API Keys.
- Python uv/uvx available on the host running the MCP client.
- Connected data sources already onboarded in AnomalyArmor.
- Claude Code, Cursor, or another MCP client with stdio transport.
Schema details
- Install type
- cli
- Reading time
- 4 min
- Difficulty score
- 40
- Troubleshooting
- Yes
- Breaking changes
- No
- Scope
- Source repo
- Estimated setup
- 10 minutes
- Difficulty
- intermediate
- Website
- https://anomalyarmor.ai
Full copyable content
{
"mcpServers": {
"armor": {
"command": "uvx",
"args": ["armor-mcp"]
}
}
}About this resource
Overview
AnomalyArmor Agents ships armor-mcp, a Model Context Protocol server for data observability. Fifty-two consolidated tools cover health summaries, alerts, freshness, schema drift, data quality, lineage, tags, and AI-driven monitoring recommendations—designed for data teams using Claude Code or Cursor.
Source lives in anomalyarmor/agents. Docs: docs.anomalyarmor.ai.
Features
- Consolidated action-parameter tools to reduce context window usage.
- Health summaries, daily briefings, and alert inbox management.
- Freshness schedules and schema drift monitoring.
- Data quality metrics, validity rules, and referential checks.
- AI recommendations and natural language ask_question intelligence.
Use Cases
- Ask whether warehouse tables are healthy this morning.
- Investigate why the orders table is stale and which dashboards are affected.
- Create freshness monitoring for high-traffic finance tables.
- Review unresolved critical alerts from yesterday.
- Get AI recommendations for tables lacking coverage.
Installation
Claude (Connectors)
- Generate an
aa_live_API key in AnomalyArmor. - Set ARMOR_API_KEY or create
~/.armor/config.yaml. - Add
uvx armor-mcpto MCP client stdio configuration. - Invoke health_summary or get_todays_briefing to verify connectivity.
Claude Code
claude mcp add armor -- uvx armor-mcp
claude mcp list
Other MCP clients
Add the connector using the JSON configuration below in your MCP client settings.
Configuration
{
"mcpServers": {
"armor": {
"command": "uvx",
"args": ["armor-mcp"]
}
}
}
For authenticated setups:
{
"mcpServers": {
"armor": {
"command": "uvx",
"args": ["armor-mcp"],
"env": {
"ARMOR_API_KEY": "aa_live_your_key_here"
}
}
}
}
Alternatively install the Claude plugin: claude plugin install armor@anomalyarmor.
Examples
Health check
Is my data healthy? Summarise alerts, stale tables, and schema changes.
Investigate stale table
Why is the orders table stale and what depends on it?
Recommend monitoring
What should I monitor on my Snowflake warehouse based on table patterns?
Security
- Treat alert rule creation as production-impacting configuration.
- Scope API keys to the minimum AnomalyArmor workspace needed.
- Review ask_question responses before sharing externally—they may quote internal schema names.
Troubleshooting
No API key configured
Set ARMOR_API_KEY or add api_key to ~/.armor/config.yaml.
MCP server not found
Verify uvx is installed and restart the MCP client after config edits.
Empty asset list
Connect a data source and run discovery in AnomalyArmor first.
Plugin skills missing
Run claude plugin list to confirm armor@anomalyarmor is installed.
Duplicate Check
No existing AnomalyArmor or armor-mcp entry was found in content/mcp/. This differs from generic SQL MCP servers because it targets observability alerts, freshness, and schema drift—not ad hoc querying.
Source citations
Add this badge to your README
How it compares
AnomalyArmor MCP Server for Claude side by side with 3 alternatives on trust, install, platform support, and disclosed safety notes — all from reviewed registry metadata.
| Field | AnomalyArmor MCP Server for Claude AnomalyArmor armor-mcp stdio server with 52 consolidated data observability tools for alerts, freshness, schema drift, quality metrics, lineage, and AI recommendations. Open dossier | Datadog MCP Server for Claude Official Datadog MCP Server for connecting Claude, Codex, Cursor, and other MCP clients to Datadog observability, incidents, logs, metrics, traces, dashboards, monitors, notebooks, services, and security signals. Open dossier | Adeu MCP Server for Claude Adeu MCP server and SDKs translate DOCX files to token-efficient Markdown with CriticMarkup, apply safe Track Changes edits, and optionally integrate with live Microsoft Word on Windows. Open dossier | Agentgateway MCP Gateway Apache-2.0 agentic proxy that can expose stdio, HTTP, SSE, and Streamable HTTP MCP servers through a managed gateway with federation, OAuth/JWT authentication, RBAC/CEL policy, CORS, TLS, observability, and Kubernetes Gateway API support. Open dossier |
|---|---|---|---|---|
| Trust | ||||
| Install risk | Review first | Review first | Review first | Review first |
| Notes | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ | Safety ✓ Privacy ✓ |
| Category | mcp | mcp | mcp | mcp |
| Source | source-backed | source-backed | source-backed | source-backed |
| Author | AnomalyArmor | Datadog | Adeu | agentgateway |
| Added | 2026-06-14 | 2026-06-04 | 2026-06-14 | 2026-06-06 |
| Platforms | Claude CodeCodexCursorClaude Desktop | Claude CodeClaude Desktop | Claude CodeClaude Desktop | Claude CodeClaude Desktop |
| Source repo | — | — | — | — |
| Safety notes | ✓Tools can create alert rules, freshness schedules, metrics, and monitoring configurations. manage_asset and related tools may trigger discovery jobs on production sources. Alert acknowledgement and resolution changes operational state. Review AI recommendations before applying automated monitoring at scale. | ✓Treat Datadog MCP Server as a production observability and operations interface. It can expose live incidents, monitors, logs, metrics, traces, dashboards, notebooks, services, RUM events, hosts, SLOs, and security context to the connected assistant. Datadog separates the main remote MCP Server from the narrower local Code Security MCP Server. Use this entry for Datadog telemetry and platform tools; use the Code Security MCP Server only when the task is explicitly local SAST, SCA, IaC, secrets, or SBOM scanning. Enable the smallest useful toolsets first. Use `toolsets` and `omit_tools` to keep context, permissions, and accidental action surface smaller than `toolsets=all`. Some tools can create or edit Datadog resources such as monitors and notebooks. Require human review for generated monitor definitions, alerting changes, incident updates, notebook edits, workflow actions, or any operation that could page a team or change production observability state. Datadog documents MCP-specific RBAC permissions as well as normal underlying resource permissions. A user who can connect the MCP Server may still need separate read or write permissions for monitors, logs, APM, dashboards, notebooks, incidents, services, and other products. The MCP Server has fair-use limits documented as 50 requests per 10 seconds for tool-call bursts and 50,000 monthly tool calls, and Datadog notes that those limits are subject to change. Datadog states that the MCP Server is under significant development. Review the official docs and available tool list before relying on a remembered tool name, permission, limit, or behavior. | ✓`process_document_batch` can modify DOCX content and apply Track Changes when edits pass validation. `finalize_document` can strip metadata and lock documents; review output before external distribution. Live Word integration edits the active document in memory; save backups before batch operations. Adeu Cloud email workflows fetch external attachments; treat counterparty documents as untrusted input. | ✓Agentgateway can aggregate and expose many downstream MCP targets through one endpoint; every target's permissions become part of the gateway surface. Stdio targets are spawned by the gateway process, so commands such as package runners or containers inherit the gateway host's trust boundary. Demo configs use permissive CORS and local authorization servers; lock down origins, headers, issuers, audiences, JWKS, and resource metadata before production use. CEL/RBAC policies can inspect MCP tool names, arguments, results, prompts, resources, request bodies, JWT claims, API keys, and backend metadata; test policies before relying on them. Remote MCP proxying, OAuth provider adaptation, Kubernetes routing, TLS termination, and guardrails are infrastructure changes that should go through normal security review. |
| Privacy notes | ✓Table names, schema metadata, alert details, and sample metrics are sent to AnomalyArmor. Natural language `ask_question` queries may expose internal data catalog context to the LLM. API keys grant access until revoked in AnomalyArmor settings. | ✓Datadog says all MCP Server tool calls are recorded in Audit Trail with MCP metadata including tool name, arguments, user identity, and MCP client. Datadog also emits MCP usage metrics such as `datadog.mcp.session.starts` and `datadog.mcp.tool.usage`, tagged with values including user ID, user email, client name, and tool name. Remote Datadog MCP Server usage can include prompts, transitions to and from the Datadog login page, errors, user identifiers, and context leading to MCP tool usage; Datadog documents that this data is stored for 120 days. Tool results can include production logs, traces, metric values, monitor states, dashboard names, notebook contents, incidents, service catalog metadata, host data, security findings, RUM events, team ownership, and operational timelines. Claude transcripts, Codex logs, IDE logs, screenshots, exported notebooks, support bundles, and incident summaries can retain Datadog-derived data outside Datadog's normal access controls and retention policies. Datadog describes the MCP Server as HIPAA-eligible, but users remain responsible for confirming that any connected AI tool, client, workflow, and retention path meets their compliance requirements. | ✓Local Adeu tools process documents on your machine and do not phone home with file contents by default. Document text sent to your LLM provider is subject to that provider's data handling policies. Optional Adeu Cloud routes large validation or email workflows through Adeu servers under separate terms. | ✓MCP requests, tool arguments, tool results, prompt names, resource names, session IDs, JWT/API-key claims, raw request/response bodies, logs, traces, and telemetry may pass through the gateway. CEL policy and observability features may buffer or inspect request and response bodies depending on configuration. OAuth/JWT metadata, bearer tokens, API keys, DCR secrets, Keycloak/Auth0/Okta settings, and upstream MCP credentials must be protected as secrets. Any data returned by downstream MCP tools can still be sent onward by the MCP client to the configured model provider. |
| Prerequisites |
|
|
|
|
| Install | | | | |
| Config | | | | |
| Citations | ||||
| Claim | Unclaimed | Unclaimed | Unclaimed | Unclaimed |
Featured in
Signals
Loading live community signals…
A short, calm digest of reviewed Claude resources. Unsubscribe any time.