# Cb Analytics Cluster

> |

- **Type:** Skill
- **Install:** `agentstack add skill-celticht32-couchbase-skills-for-claude-ai-cb-analytics-cluster`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [celticht32](https://agentstack.voostack.com/s/celticht32)
- **Installs:** 0
- **Category:** [Databases](https://agentstack.voostack.com/c/databases)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [celticht32](https://github.com/celticht32)
- **Source:** https://github.com/celticht32/Couchbase-Skills-for-Claude.ai/tree/main/skills/couchbase-analytics/cb-analytics-cluster

## Install

```sh
agentstack add skill-celticht32-couchbase-skills-for-claude-ai-cb-analytics-cluster
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Cluster operations

You have 9 cluster-level tools, mostly read-only.

## Reachability and identity

- `ping_cluster(cluster)` — fast yes/no liveness check. Use this first when
  troubleshooting any other tool failure.
- `get_cluster_info(cluster)` — UUID + implementation version. Cheap; safe
  to call often.
- `who_am_i(cluster)` — what the cluster sees the current MCP user as
  (roles, domain). Useful when an `AnalyticsAuthError` shows up.

## Capacity and health

- `get_cluster_details(cluster)` — nodes, memory quotas (`memoryQuota`,
  `cbasMemoryQuota`), `balanced`, `rebalanceStatus`. Read before any
  capacity decision.
- `get_cluster_tasks(cluster)` — every in-flight task (rebalance, compaction,
  XDCR). Empty list is the happy path.
- `get_rebalance_progress(cluster)` — focused view: rebalance only.
  Returns `{"status": "none"}` when nothing's running.

## Auto-failover

- `get_auto_failover_settings(cluster)` — read current settings.
- `configure_auto_failover(enabled, timeout, max_count, cluster)` —
  Couchbase recommends `enabled=True, timeout=120, max_count=1` for
  3-node clusters; larger clusters can use `max_count=2+`.

## System events

`get_system_events(since_time, cluster)` returns recent operational events
(`info` / `warning` / `error`). `since_time` is an ISO 8601 timestamp; omit
it for "everything the cluster will give us".

Use this for post-incident triage:

> User: "Why did the cluster fail over at 03:00?"
> 1. `get_system_events(since_time="2026-05-24T02:50:00Z")`
> 2. Look for `severity="warning"|"error"` entries around that timestamp.

## Confirmation patterns

- `configure_auto_failover` *changes durable cluster state*. Always confirm
  before calling. Restate the values you're about to set and which cluster.
- Never call any of these against the wrong cluster — `cluster` parameter
  is optional but if you have any doubt, pass it explicitly. Use
  `list_clusters` from the meta tools when in doubt.

## What to avoid

- Don't poll `get_cluster_tasks` faster than once a second during a long
  rebalance; the management endpoint is shared with the GUI.
- Don't disable auto-failover in production unless you have an immediate
  reason and a plan to re-enable. Note the original values in chat before
  you change them so they're easy to restore.

## Rate limits & safety

Cluster tools are almost all `read` (60/sec):
`ping_cluster`, `get_cluster_info`, `get_cluster_details`,
`get_cluster_tasks`, `get_rebalance_progress`,
`get_auto_failover_settings`, `get_system_events`, `who_am_i`.

The single `write` (1/sec) tool here is `configure_auto_failover` —
matches the safety advice above: never call this without explicit
confirmation, and the rate limit gives you exactly one shot per second
to fat-finger it.

When polling `get_rebalance_progress` during a long rebalance, the read
rate (60/sec) is plenty but the management endpoint is shared with the
GUI. Once every 2–5 seconds is the polite poll interval; faster won't
get you better data and stresses the management plane.

If `RateLimitExceeded` comes back, honour `retry_after_sec` — back off,
don't retry-storm.

## Related skills

- `cb-analytics-admin` — Analytics service-level health (ingestion status, active queries) rather than cluster-level
- `couchbase-mcp` — full cluster administration (rebalance, node management, XDCR, eventing) via MCP-Couchbase

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [celticht32](https://github.com/celticht32)
- **Source:** [celticht32/Couchbase-Skills-for-Claude.ai](https://github.com/celticht32/Couchbase-Skills-for-Claude.ai)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-celticht32-couchbase-skills-for-claude-ai-cb-analytics-cluster
- Seller: https://agentstack.voostack.com/s/celticht32
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.