N-able Skills Library para Claude Code
theonlytruebigmac/nable-skills
Skills MSP para gerenciar N-able/N-central com GraphQL e REST
Ver no GitHubBiblioteca de skills MSP para Claude Code integrando as APIs N-able e N-central. Oferece 44+ skills prontos para gerenciar vulnerabilidades, alertas, tickets e monitoramento em tempo real via GraphQL (N-able MCP) e REST (N-central MCP) de forma complementar.
Instalação
npx convez-ai theonlytruebigmac/nable-skills1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Triage automático de alertas de segurança e vulnerabilidades
- Gerenciamento de tickets PSA e manutenção de infraestrutura
- Monitoramento e análise de métricas em tempo real
- Consultas híbridas combinando dados de GraphQL e REST
Veredito

Solução robusta e bem arquitetada para MSPs que usam N-able/N-central. Requer setup de MCPs e compreensão das diferenças entre as duas APIs, mas compensa com cobertura completa e skills hybrid poderosos.
README
A tradução para o português deste README ainda está na fila.
N-able / N-central MSP Skill Library
MSP skill library for Claude Code, covering two N-able/N-central APIs:
- N-able MCP (
nable-mcp) — the official GraphQL preview endpoint (n-query based). - N-central MCP (
N-central-mcp) — the classic N-central REST API.
The two are complementary: the GraphQL API is strong on vulnerabilities, per-patch detail, and metrics; the classic REST API is strong on live monitoring alerts, PSA/ticketing, custom properties, maintenance windows, warranty/lifecycle, RBAC, and deployment. A set of hybrid skills uses both at once.
📖 Which MCP should I use? · 📖 N-central MCP tool reference
Install
This repo is a Claude Code plugin marketplace. Install all skills with:
/plugin marketplace add theonlytruebigmac/nable-skills
/plugin install ncentral@nable-skills
Skills are then available namespaced as /ncentral:<skill-name> (e.g. /ncentral:alert-triage). Pull updates later with /plugin marketplace update nable-skills.
The skills call the N-able and/or N-central MCP servers — but this plugin ships skills only, so you connect an MCP yourself. See Connecting an MCP.
Repository layout
This repo is a marketplace (catalog) with one plugin per N-able product. Today that's ncentral; future products (e.g. N-sight, Cove) become sibling plugins under plugins/, each installable on its own.
nable-skills/ ← marketplace (catalog)
├── .claude-plugin/marketplace.json
├── docs/ ← shared reference
└── plugins/
└── ncentral/ ← the N-central product plugin
├── .claude-plugin/plugin.json
└── skills/ ← all 44 N-central skills
Connecting an MCP
This plugin ships skills only — it does not bundle or configure any MCP server. The skills call tools on the N-able / N-central MCPs, so connect at least one of them yourself; the skills then use whatever's available. An unconnected MCP just means the skills that need it can't run — the rest still work (hybrid-* skills need both).
| MCP | What it is | How to set it up |
|---|---|---|
nable-mcp | N-able's official hosted GraphQL MCP (covers N-central + N-sight) | Docs: developer.n-able.com/gql/docs/mcp. Generate a token at n-able.app/api-token-management. |
N-central-mcp | Classic N-central REST MCP — deeper REST coverage | theonlytruebigmac/n-central-mcp; point it at your N-central with a User-API JWT. |
Where the config goes depends on the surface:
- Claude Desktop → Chat / Cowork:
~/Library/Application Support/Claude/claude_desktop_config.json(mcpServers). - Claude Code:
~/.claude/mcp.json, or a project-level.mcp.json.
Minimal nable-mcp example (drop into the mcpServers block of either file):
{
"mcpServers": {
"nable-mcp": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://api.n-able.com/mcp", "--header", "Authorization: Bearer YOUR_NABLE_TOKEN"]
}
}
}
In Claude Code, run /mcp to check connection status.
MCP tools
N-able MCP (GraphQL)
| Tool | Purpose |
|---|---|
introspect | Discover fields on any schema type |
search | Find queryable fields by keyword |
validate | Check a query before running — always first |
execute | Run a GraphQL query against your fleet |
get_asset_inventory | Paginated list of managed devices |
get_single_asset_details_and_metrics | Full detail + CPU/memory for one device |
list_script_tasks_for_asset | Script execution history for a device |
assign_tags_to_assets | Bulk-tag devices (write) |
Rule: never execute without passing validate first.
Confirmed GraphQL queries (used by current skills): assetSearch, asset(id),
assetAggregations, patchInstallationSearch, patchInstallationAggregations,
vulnerabilityDetectionSearch, vulnerabilityDetectionAggregations,
organizationSearch, taskSearch, taskAggregations, tagSearch. The asset(id)
node also exposes nested activitySearch and taskExecutionSearch (used by
incident-summary).
N-central MCP (classic REST)
~80 tools spanning devices, live monitoring (list_active_issues, get_device_status),
scheduled tasks, custom properties, maintenance windows, warranty/lifecycle, PSA/ticketing,
users/roles/access, device notes, deployment, reports, and org provisioning. See the full
catalog and call conventions in docs/ncentral-mcp-reference.md.
Org hierarchy
N-able GraphQL N-central REST
Partner Service Organization (soId)
└── ServiceOrganization └── Customer (customerId)
└── Customer └── Site (siteId)
└── Site └── Device (deviceId)
└── Asset
GraphQL assetSearch accepts inOrganizations: [id] at any level. N-central uses numeric
orgUnitId/soId/customerId/siteId and string deviceId.
⚠ The two ID spaces are different. Never pass a GraphQL id to N-central or vice versa — join by customer name and device name/hostname. See Cross-MCP correlation.
Skills
N-able MCP — GraphQL skills
Engineer
| Skill | What it does | Cadence |
|---|---|---|
alert-triage | Disconnected agents + patch errors + critical vulns | Daily / on-call |
patch-status | Patch compliance by status across customer or fleet | Weekly |
patch-approval-queue | AVAILABLE and APPROVED patches awaiting action | Daily / patch Tuesday |
device-health-check | Full device snapshot with metrics | Before any touch |
onboarding-checklist | Gap analysis vs. MSP baseline | After agent deploy |
incident-summary | Timeline from activity logs + task history | Post-incident |
stale-device-audit | Disconnected devices + license cleanup | Monthly |
automation-policy-review | Script task failures and coverage | Monthly |
eol-os-report | EOL OS inventory with upgrade priority | Quarterly |
network-discovery-review | Unmanaged and disconnected device gaps | Monthly |
site-dispatch-brief | Pre-on-site briefing for field techs | Before every visit |
tag-devices | Assign tags in bulk (write) | On demand |
snapshot | Save point-in-time fleet state for trends | Weekly / before QBR |
Executive / Sales
| Skill | What it does | Cadence |
|---|---|---|
qbr-brief | Fleet + patch + vuln summary for QBR | 1 week before QBR |
health-score | Weighted 0–100 environment health score | Monthly |
renewal-risk-report | Customers with deteriorating signals | Weekly |
upsell-opportunity-report | Coverage gaps mapped to service offerings | Monthly |
executive-monthly-digest | Full-fleet KPI memo for leadership | 1st of month |
new-prospect-baseline | Pre-sales scoping assessment | Before SOW |
sla-compliance-report | Patch/vuln age vs. SLA thresholds | Monthly |
fleet-growth-report | Device count and composition by customer | Monthly |
N-central MCP — classic REST skills
Monitoring & ops
| Skill | What it does | Cadence |
|---|---|---|
active-issues | Live monitoring-alert triage (the real alerts) | Daily / on-call |
scheduled-tasks | Scheduled-task & job-status monitor | Daily / weekly |
maintenance-windows | Review/create/update patch windows (write) | Monthly / onboarding |
Asset & lifecycle
| Skill | What it does | Cadence |
|---|---|---|
warranty-lifecycle | Warranty + hardware-refresh forecast & budget | Quarterly |
custom-property-audit | MSP metadata completeness + bulk fill (write) | Monthly |
device-notes | Read/add/bulk device runbook notes (write) | On demand |
asset-inventory | Hardware/software asset inventory | Quarterly |
Admin & onboarding
| Skill | What it does | Cadence |
|---|---|---|
psa-ticketing | PSA tickets + customer↔company mapping (write) | On demand |
deployment-kit | Tokens/installers/keys to deploy the agent | On onboarding |
license-capacity | License limits vs. usage / headroom | Monthly |
access-review | Users, roles, access groups audit | Quarterly |
org-hierarchy | SO→Customer→Site export + ID lookup | On demand |
patch-comparison | N-central native patch comparison report | Weekly |
fleet-export | Full spreadsheet-ready device export | On demand |
server-health | N-central server version/uptime/clock drift | Weekly |
Hybrid skills — both MCPs (hybrid-*)
| Skill | What it does | Cadence |
|---|---|---|
hybrid-device-360 | One device's complete picture across both APIs | Before any touch |
hybrid-alert-triage | N-central live alerts ranked by GraphQL CVE/patch risk | Daily / on-call |
hybrid-customer-360 | Full account rollup: ops + business signals | Before QBR / escalation |
hybrid-health-score-plus | health-score + monitoring, warranty, metadata, license | Monthly |
hybrid-onboarding-plus | onboarding-checklist + N-central config completeness | After onboarding |
hybrid-incident-bridge | Incident timeline from both, then draft a PSA ticket | Post-incident |
hybrid-qbr-plus | qbr-brief + hardware refresh + tickets + license | 1 week before QBR |
hybrid-patch-reconciliation | Unpatched devices × maintenance windows = root cause | Weekly / patch Tuesday |
Meta / authoring
| Skill | What it does | Cadence |
|---|---|---|
create-skill | Token-minimal template, token-cost model, and review checklist for authoring a new skill in this library | On demand |
Gaps & guidance
- Live alerts live in N-central (
list_active_issues), not GraphQL — use theactive-issues/hybrid-alert-triageskills for real alert state. - CVE vulnerabilities, tags, and CPU/memory metrics live in GraphQL, not REST.
- Month-over-month deltas: both APIs are current-state — use
snapshotto build trends. - Ticket response times live in your PSA; N-central can create/read Custom PSA tickets
and mappings via
psa-ticketing. - Writes (tagging, maintenance windows, custom properties, notes, lifecycle, tickets, provisioning) always preview and confirm before executing.