{"id":"a1ae11c8-2dc4-4a8f-bd31-78b4dc44e27e","slug":"ceph-cookbook","kind":"mcp","name":"ceph-cookbook","summary":"Claude Code plugin for deploying, operating, and troubleshooting Ceph clusters (Rook & cephadm) — skills, agents, hooks, and monitors for day-2 ops.","owner":"sserkanml","repoUrl":"https://github.com/sserkanml/ceph-cookbook","license":"GPL-3.0","licenseTier":"strong-copyleft","trust":"community","stars":2,"match":null,"fewStars":true,"format":"mcp","createdAt":"2026-07-14T18:15:49.000Z","addedAt":"2026-10-21T09:18:08.735Z","logoUrl":"https://avatars.githubusercontent.com/u/86930300?u=99cda89b89fba1a6055949d4d7cad5066a2f4844&v=4","linkOnly":false,"collection":null,"skillPath":null,"categories":["developer-tools","cloud-infra"],"tags":["ceph","claude-code","devops"],"runtime":"binary","local":true,"cloudCapable":false,"cloudCapability":"unknown","cloudReadyOnPlatform":false,"needsApiKey":false,"reviewed":false,"syncedAt":"2026-10-23T01:52:10.194Z","sample":false,"url":"https://conduits.dev/tools/ceph-cookbook","language":"Shell","transports":["stdio"],"install":{},"requirements":{"env":[],"upstreamMayCharge":false},"tools":null,"readme":"# ceph-cookbook\n\nA Claude Code plugin for deploying, operating, and troubleshooting Ceph storage clusters —\nskills, agents, hooks, and monitors covering cluster setup (Rook or cephadm), block/file/\nobject storage, day-2 lifecycle operations, security, and observability.\n\n## Installation\n\nAdd this repo as a plugin marketplace, then install the plugin:\n\n```\n/plugin marketplace add sserkanml/ceph-cookbook\n/plugin install ceph-cookbook\n```\n\n## Skills\n\n### Cluster deployment\n| Skill | Use for |\n|---|---|\n| `ceph-deploy-rook` | Bootstrap a Ceph cluster on Kubernetes via the Rook operator + `CephCluster` CRD |\n| `ceph-deploy-cephadm` | Bootstrap a Ceph cluster on bare-metal/VMs via `cephadm` + `ceph orch` |\n\n### Object storage (RGW)\n| Skill | Use for |\n|---|---|\n| `rgw-multisite-ceph` | RGW multisite replication between two cephadm-managed clusters (`radosgw-admin` + `ceph orch`) |\n| `rgw-multisite-rook` | RGW multisite replication between two Rook-managed clusters (`CephObjectRealm`/`CephObjectZone` CRDs) |\n| `rgw-user-quota-mgmt` | RGW S3 user lifecycle, access key rotation, quotas, bucket policies |\n\n### Block & file storage\n| Skill | Use for |\n|---|---|\n| `rbd-pool-setup` | Create an RBD pool/image, scope a cephx client, map/mount |\n| `rbd-mirror-dr` | RBD mirroring (journal or snapshot mode) between two clusters for block-storage DR |\n| `cephfs-deploy` | Deploy MDS daemons and create a CephFS filesystem via cephadm |\n| `nfs-ganesha-export` | Export CephFS or an RGW bucket over NFS via `ceph orch`-managed NFS-Ganesha |\n\n### Cluster lifecycle & tuning\n| Skill | Use for |\n|---|---|\n| `ceph-upgrade` | Rolling cluster upgrade via `ceph orch upgrade`, with health gating |\n| `osd-host-lifecycle` | Add/drain/remove OSDs and hosts, replace a failed disk |\n| `crush-pool-tuning` | Replicated vs. erasure-coded pools, CRUSH failure domains, PG autoscaling |\n\n### Security & observability\n| Skill | Use for |\n|---|---|\n| `cephx-key-management` | Least-privilege client keyrings, capability auditing, key rotation/revocation |\n| `ceph-prometheus-grafana` | Enable the mgr `prometheus` module and wire up Prometheus/Grafana dashboards |\n\n## Agents\n\n| Agent | Use for |\n|---|---|\n| `ceph-diagnostician` | Read-only health/OSD/PG/RGW-sync diagnosis; reports issues and recommended next steps, never remediates |\n| `ceph-remediator` | Executes a fix for an already-diagnosed issue, with explicit risk classification and confirmation before medium/high-risk actions |\n\n## Hooks & Monitors\n\n- **`guard-dangerous-commands.sh`** (`PreToolUse` hook) — blocks irreversible commands\n  (pool delete, OSD purge, realm/zone delete, force-repair, `--yes-i-really-mean-it`)\n  before they run, regardless of prior confirmation.\n- **`health-watch.sh`** (monitor) — reports `HEALTH_OK`/`WARN`/`ERR` transitions.\n- **`rgw-sync-watch.sh`** (monitor) — reports RGW multisite sync status transitions\n  (caught up / behind / failed).\n\n## Output Styles\n\n| Style | Use for |\n|---|---|\n| `ceph-incident` | Fast, command-first responses during a live incident |\n| `ceph-postmortem` | Root-cause analysis and prevention, after the fact |\n| `ceph-runbook` | Step-by-step, verifiable procedures for planned maintenance/setup |\n\n## License\n\nGPL-3.0 — see [LICENSE](LICENSE).\n"}