Skip to main content
Glama
Bitdefender GravityZone MCP Server

Bitdefender GravityZone MCP Server

by WYRE-AI

Bitdefender GravityZone MCP Server

MCP server for Bitdefender GravityZone's Control Center API - endpoint/network inventory, security policies, the account-wide malware-hash Blocklist, and scheduled/instant reports - for AI assistants and the WYRE Conduit gateway.

New call shape for this catalog: GravityZone's API is JSON-RPC 2.0, not REST - every other Conduit sidecar (Cork, CyberQP, Slide, UniFi, Cisco Duo) calls a conventional REST API. See Authentication and JSON-RPC shape below.

Authentication

GravityZone authenticates with HTTP Basic Authentication (RFC 2617): the API key is sent as the username, with an empty password - Authorization: Basic base64(apiKey + ":"). This is also a new shape for this catalog - every other static-key connector here (Cork, CyberQP, Slide, UniFi) sends a custom X-Vendor-Api-Key header that the sidecar turns into a Bearer token; GravityZone has no bearer-token concept at all.

API keys are generated in Control Center under My Account -> API keys, and each key is scoped, at generation time, to a checklist of API areas (Network, Policies, Reports, Incidents, Companies, Licensing, Accounts, Packages, Integrations, Push, Quarantine). This connector's key needs the Network, Policies, Reports, Incidents, and Quarantine areas selected - the other six are never called and do not need to be enabled. If the current key was generated without Quarantine, the quarantine tool returns HTTP 403 until a key with it is in place. bitdefender_list_incidents / bitdefender_get_incident also need a GravityZone license that includes incidents (EDR/XDR) - without one they return a JSON-RPC error.

This connector also needs the account's Control Center Access URL (shown on that same My Account page, under Control Center API) - GravityZone Cloud, its EU region, and any self-hosted/on-premise Control Center each have a different API host, so unlike most vendors in this catalog there is no single fixed base URL to hardcode.

In gateway mode, both arrive per-request via the X-Bitdefender-Api-Key and X-Bitdefender-Access-Url headers; in local/stdio mode they're read once from BITDEFENDER_API_KEY and BITDEFENDER_ACCESS_URL.

JSON-RPC shape

GravityZone exposes several separate JSON-RPC 2.0 endpoints, one per product area, at <accessUrl>/<version>/jsonrpc/<area> (e.g. /v1.0/jsonrpc/network, /v1.0/jsonrpc/policies). Most methods are under v1.0; the incident-retrieval methods are only under v1.2 (/v1.2/jsonrpc/incidents), and quarantine has a service suffix (/v1.0/jsonrpc/quarantine/computers). Every call to a given area - whatever it does - is an HTTP POST to that same URL with an identical envelope:

{ "id": "<uuid>", "jsonrpc": "2.0", "method": "getEndpointsList", "params": { "parentId": "..." } }

The operation lives entirely inside the body (method + params), never in the URL or HTTP verb - there is no GET/POST/PATCH/DELETE distinction to key a read-only boundary off of, unlike every REST sidecar in this catalog. This connector's read-only scope is therefore enforced by only ever calling documented getXxx/listXxx method names (see Scope below and client.ts's jsonRpcCall()), not by restricting HTTP verbs.

A successful call returns { "id", "jsonrpc", "result": ... }. A failed call can come back two ways, and this connector's client.ts distinguishes both:

  • HTTP-layer failure - 401 (bad/missing key), 403 (the key doesn't have this API area enabled - see above), 405 (non-POST), or 429 (rate limit).

  • JSON-RPC-layer failure - HTTP 200 with a body error member ({code, message, data.details}, e.g. -32602 Invalid params).

Rate limit: 10 requests/second per API key; GravityZone returns 429 above that.

Related MCP server: ESET PROTECT MCP Server

Configuration

Env var

Description

BITDEFENDER_API_KEY

API key generated in Control Center, with the Network/Policies/Reports/Incidents/Quarantine areas selected.

BITDEFENDER_ACCESS_URL

The account's Control Center Access URL (e.g. https://cloud.gravityzone.bitdefender.com).

MCP_TRANSPORT

stdio (default) or http.

AUTH_MODE

env (default, reads the two vars above) or gateway (credentials arrive per-request via the X-Bitdefender-Api-Key / X-Bitdefender-Access-Url headers, injected by the Conduit gateway).

CONDUIT_S2S_SECRET

When set, the HTTP transport requires a valid X-Gateway-S2S header (Conduit sidecar auth) on every /mcp request.

LOG_LEVEL

debug | info (default) | warn | error.

Request limits

  • BITDEFENDER_MAX_UPSTREAM_BYTES caps each GravityZone response at 16 MiB (16777216 bytes) by default, enforced while streaming before JSON parsing. Positive values are capped at 128 MiB (134217728 bytes); missing, blank, invalid, or non-positive values fall back to the default.

  • HTTP request bodies are capped at 256 KiB (262144 bytes). After the X-Gateway-S2S check passes, a Content-Length above the cap returns 413 with a JSON-RPC error; bodies without Content-Length are capped by the MCP SDK transport. Unauthenticated requests still get 401 first.

  • tools/call arguments are capped at 10,000 elements (every array item and object key at any depth). Larger inputs return an isError tool result and are never dispatched.

These limits address CVE-2026-111768 / GHSA-46jr-4wx8-v453 (memory exhaustion during validation) and GHSA-f897-62hf-6gxq (no request body / batch cap).

Tools

Network - endpoint/network inventory

  • bitdefender_list_endpoints - list managed/unmanaged endpoints (name, FQDN, IP, MACs, group, agent flags), optionally scoped and filtered.

  • bitdefender_get_endpoint - full detail for one managed endpoint (agent version, license status, scan/update status).

  • bitdefender_list_custom_groups - list the child groups under a Network Inventory group.

  • bitdefender_list_network_inventory - list inventory items (groups, computers, VMs, EC2 instances) with per-type filtering.

  • bitdefender_list_scan_tasks - list previously created on-demand scan tasks and their status.

Policies

  • bitdefender_list_policies - list security policies available to the account.

  • bitdefender_get_policy - get full settings for one security policy.

Incidents

  • bitdefender_list_blocklist_items - list file hashes present in the account's Blocklist.

  • bitdefender_list_incidents - list EDR/XDR incidents (status, priority, main action, timestamps), filterable by endpoint, status, priority, type and date. Calls getIncidentsList on Incidents API v1.2 (<accessUrl>/v1.2/jsonrpc/incidents); needs a license that includes incidents.

  • bitdefender_get_incident - full detail for one incident (severity, attack types, affected computer, nodes, alerts, MITRE tags). Calls getIncident on Incidents API v1.2.

Quarantine

  • bitdefender_list_quarantine_items - files quarantined on computers and VMs (threat name, path, SHA256, endpoint, date), for one endpoint or the whole network. Calls getQuarantineItemsList on <accessUrl>/v1.0/jsonrpc/quarantine/computers. The Exchange quarantine service is not called.

Reports

  • bitdefender_list_reports - list scheduled/instant reports configured on the account.

  • bitdefender_get_report_download_links - check download readiness and get a report's download URL(s).

Scope

This is a deliberately narrow, read-only surface: exactly 13 of GravityZone's getXxx/listXxx methods across the Network, Policies, Incidents, Quarantine, and Reports APIs. Every tool is classified isAdmin: true in the Conduit gateway - GravityZone is an endpoint-protection product, so every read this connector exposes (endpoint IP/MAC/hostname inventory, security policy settings, malware-hash blocklist entries, security posture reports) is security/PII-adjacent by construction. No mutating (create/update/delete/move/set/add/remove) method is implemented, by design, not by oversight:

Hard-excluded (Network API) - never implemented:

  • createCustomGroup, deleteCustomGroup, moveCustomGroup - group provisioning/mutation.

  • moveEndpoints, deleteEndpoint - endpoint mutation/removal.

  • createScanTask - dispatches a real on-demand scan to a real managed endpoint. bitdefender_list_scan_tasks (read) stays; this does not.

  • setEndpointLabel - endpoint mutation.

Hard-excluded (Incidents API) - never implemented:

  • addToBlocklist, removeFromBlocklist - Blocklist mutation. bitdefender_list_blocklist_items (read) stays; these do not.

  • createIsolateEndpointTask, createRestoreEndpointFromIsolationTask - genuine remote-response actions that isolate/restore a real managed endpoint from the network.

  • changeIncidentStatus, changeIncidentPriority, assignIncident, updateIncidentNote - incident triage mutation. bitdefender_list_incidents / bitdefender_get_incident (read) stay; these do not.

  • createResponseAction, startYaraScan, createCustomRule, updateCustomRule, deleteCustomRule - response actions and detection-rule changes.

Hard-excluded (Quarantine API) - never implemented:

  • createRemoveQuarantineItemTask, createRestoreQuarantineItemTask (and their Exchange equivalents), createEmptyQuarantineTask, createAddFileToQuarantineTask, and createReleaseQuarantineExchangeItemTask - write methods that empty quarantine, add, delete, or release a quarantined file on a real endpoint. bitdefender_list_quarantine_items (read) stays; these do not.

Hard-excluded (Reports API) - never implemented:

  • createReport - creates a new scheduled/instant report definition.

  • deleteReport - deletes a report definition.

Out of scope entirely (not requested, not implemented): the Companies, Licensing, Accounts, Packages, Integrations, and Push APIs. Push (setPushEventSettings) would stream events to a receiver, but configuring it is itself a write, so detections are read here by pulling incidents and quarantine instead.

They can be added as a follow-up if there's demand, after a deliberate scope decision - not by default.

Credential scope

Two-tier confidence split, per this catalog's convention for a claim about what a credential can and cannot reach:

  • STRUCTURALLY VERIFIED - this connector's own code never calls a mutating JSON-RPC method: confirmed against client.ts/tools/*.ts, only the 13 documented getXxx/listXxx methods listed under Tools are ever sent, no passthrough/arbitrary-method call exists anywhere in src/. This connector is read-only by construction.

  • VENDOR-DOCUMENTED, AND THE FINDING RUNS THE OTHER WAY - GravityZone's own API-key generation UI (Control Center -> My Account -> API keys) only gates access per API area (Network/Policies/Reports/Incidents/...), confirmed against Bitdefender's official API Guide (Getting Started section 1.3, "API Keys": "Each API key allows the application to call methods exposed by one or several APIs. The allowed APIs are selected at the time the API key is generated"). There is no documented per-method or read/write permission flag. A key with the Network area enabled can call getEndpointsList (read) and moveEndpoints/deleteEndpoint/createScanTask (write) - the same area, the same permission checkbox, no finer-grained scoping exists on Bitdefender's side. This connector's read-only posture is enforced entirely by its own code, not by any narrower credential scope GravityZone itself offers - the same posture this catalog's other security-product connectors (Cork, CyberQP, Cisco Duo) already document for their own admin-tier classification, stated once here for the credential-scope question specifically.

Development

npm install
npm run build
npm test
npm run lint   # tsc --noEmit

Docker

docker build -t bitdefender-mcp .
docker run -p 8080:8080 -e BITDEFENDER_API_KEY=... -e BITDEFENDER_ACCESS_URL=... bitdefender-mcp

Available Tools

13 tools
bitdefender_get_endpointA

Get full detail for one managed endpoint: power state, IP, last-seen timestamp, and the installed security agent's engine/product versions, license status, and update/outdated flags. Requires an endpoint ID from bitdefender_list_endpoints.

Inputs: endpointId (required).

bitdefender_get_incidentA

Get full detail for one incident: severity score (1-100), priority, attack types (e.g. Malware, Ransomware, Password Stealer), detection name, affected computer (name, FQDN, IP, MACs), process/file nodes, alerts, MITRE tags, status history and the Control Center link. Requires an incident ID from bitdefender_list_incidents. Read-only.

Inputs: id (required).

bitdefender_get_policyA

Get full detail for one security policy: creator, created/last-modified dates, and its complete settings object. Requires a policy ID from bitdefender_list_policies.

Inputs: policyId (required).

bitdefender_list_blocklist_itemsA

List file hashes present in the account's Blocklist - hash, hash algorithm (SHA256/MD5), source (incident/import/manual), source description, and filename where known. Read-only - this connector never adds to or removes from the Blocklist (addToBlocklist/removeFromBlocklist are excluded; see README's Scope section). Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, perPage.

bitdefender_list_custom_groupsA

List the child groups directly under a group in the Network Inventory tree. Omit parentId to list the top-level 'Computers and Groups' and 'Deleted' groups. Group IDs from this response can be passed as parentId to bitdefender_list_endpoints and bitdefender_list_network_inventory. Returns a plain array (not a paginated envelope), each entry {id, name}.

Inputs: parentId.

bitdefender_list_endpointsA

List managed/unmanaged endpoints (computers, VMs, EC2 instances) - name, FQDN, IP, MAC addresses, AD SID, group, and installed-agent flags. Endpoint IDs from this response are required by bitdefender_get_endpoint. Scope to a group with parentId (from bitdefender_list_custom_groups); omit for the root of the Network Inventory. Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, filters, perPage, parentId, +1 more.

bitdefender_list_incidentsA

List EDR/XDR incidents (endpoint and organization incidents) - incidentId, incidentNumber, type, company, status, main action (reported/blocked/partially_blocked), priority and timestamps. Incident IDs from this response are required by bitdefender_get_incident. Scope to one machine with filters.endpointId (from bitdefender_list_endpoints). Needs a GravityZone license that includes incidents and an API key with the Incidents API enabled. Read-only - status, priority, notes and assignment are never changed. Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, sortBy, filters, perPage.

bitdefender_list_network_inventoryA

List Network Inventory items (groups, computers, virtual machines, EC2 instances) under a container, with per-type detail. Broader than bitdefender_list_endpoints, which returns only endpoints - this also returns group nodes and lets you filter by item type. Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, filters, perPage, parentId.

bitdefender_list_policiesA

List security policies available to the account (id, name, owning company). Policy IDs from this response are required by bitdefender_get_policy. Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, perPage.

bitdefender_list_quarantine_itemsA

List files quarantined on computers and virtual machines - threat name, file path, file SHA256, endpoint name/IP/ID, quarantine date and pending action status. This is the record of what the antimalware engine actually caught, including detections already cleaned. Scope to one machine with endpointId (from bitdefender_list_endpoints), or omit it for the whole network of the key's company. Needs an API key with the Quarantine API enabled. Read-only - never removes or restores a quarantined file (createRemove/createRestore quarantine tasks are excluded). Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: page, filters, perPage, endpointId.

bitdefender_list_reportsA

List scheduled/instant reports configured on the account - id, name, report type, and run occurrence (hourly/daily/weekly/monthly). Report IDs from this response are required by bitdefender_get_report_download_links. Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: name, page, type, perPage.

bitdefender_list_scan_tasksA

List previously created on-demand scan tasks and their status (1=Pending, 2=In progress, 3=Finished). Read-only - this connector never creates a scan task (createScanTask is excluded; see README's Scope section). Returns a paginated envelope: page/pagesCount/perPage/total/items.

Inputs: name, page, status, perPage.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A4.2/5.0

    Scored across 13 tools

    Disambiguation4/5

    Each tool has a clear resource+action target, and the list/get pairs (incidents, policies, endpoints) are unambiguous. The only mild overlap is the endpoints / network_inventory / custom_groups triangle, but the descriptions explicitly state the difference (network_inventory returns group nodes and supports type filtering, custom_groups returns a plain array).

    Naming Consistency5/5

    All 13 tools follow a uniform bitdefender_<verb>_<noun> snake_case pattern, with verbs consistently drawn from list/get. No camelCase mixing, no vague verbs, and the server prefix is applied uniformly.

    Tool Count5/5

    13 tools sits squarely in the well-scoped range for a security-console API surface. Each tool maps to a distinct GravityZone resource or report artifact, and no tool looks redundant.

    Completeness4/5

    The read-only scope is explicit, so the absence of create/update/delete is by design rather than a gap, and the main resources all have list+get coverage. Minor gaps remain: quarantine items, blocklist items and scan tasks have list but no detail-retrieval counterpart, and there is no account/license summary tool.

    Maintenance

    ActivityNo data
    ResponsivenessOutdated data

    Related MCP Connectors

    • Deltabridge brings an AI-native cyber threat intelligence platform into your AI agent. Search a knowledge graph of 400,000+ threat actors, malware families, vulnerabilities, indicators and campaigns, mapped to MITRE ATT&CK. Look up and enrich IOCs, trace links between entities and read living briefs built from 300+ open sources. Read-only, scoped to your organisation, with OAuth sign-in or an API key. Included on Researcher, Business and Enterprise plans.

      https://mcp.deltabridge.ai
    • Query your team's drift, vulnerability, and upgrade data from any AI assistant. OAuth 2.1, 51 tools.

      https://mcp.vibgrate.com
    • Real-time CVE, exploit, and vulnerability intelligence for AI assistants (350K+ CVEs, 115K+ PoCs)

      https://mcp.exploit-intel.com/mcp
    • Arsen is a human risk management platform that enables security teams to assess employee exposure through advanced phishing, voice phishing, and deepfake simulations, and strengthen their resilience with adaptive cybersecurity awareness training. Arsen MCP connects your Arsen platform to your preferred AI assistant, including Claude, ChatGPT, and any MCP-compatible assistant. With Arsen MCP, you can: 1. Build and launch campaigns. Let your AI assistant prepare phishing simulation campaigns and cybersecurity awareness programs, then launch them with your approval. 2. Get actionable insights from your data. Ask questions in plain language and get answers based on your organization's actual data, including campaign results, click, compromise and reporting rates, phishing score trends, coverage, and teams that need attention. 3. Manage multiple client organizations. Give MSSPs the ability to query and work across every client organization they manage, from a single AI assistant. Secure by design. Users authenticate with their Arsen account (SSO and MFA supported), and every request is governed by their existing Arsen permissions.

      https://api.arsen.co/mcp

    Related MCP Servers

    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI assistants to query and manage OpenCTI threat intelligence data, including indicators, observables, reports, malware, and more, with read-only and optional write operations.
      1
      MIT
    • A
      license
      B
      quality
      B
      maintenance
      Enables AI assistants to gain read-only visibility into an ESET PROTECT fleet via the ESET Connect API, covering devices, detections/EDR, vulnerabilities, patch status, and users.
      26
      Apache 2.0
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI assistants to query RoboShadow's Rubicon Platform Data API for read-only endpoint security, asset inventory, applications, users, vulnerabilities, and remediation status data.
      40
      1
      Apache 2.0
    • A
      license
      Not graded
      quality
      B
      maintenance
      Enables AI agents to ask Purple AI hunting questions, run PowerQueries over the Singularity Data Lake, and read alerts, vulnerabilities, misconfigurations and asset inventory from a SentinelOne console, plus VirusTotal and public CVE lookups. It is read-only and accepts per-request console credentials, making it stateless and multi-tenant.
      Apache 2.0