Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Manage OceanBase tenants using obd. Create, drop, show, and optimize tenants. Configure backup paths, run backups, and restore from backup. Use when users ask about OceanBase tenants, tenant creation, backup, restore, workload optimization, or multi-tenant management via obd.
.claude/skills/oceanbase-tenant-management/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 55% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 73% | 0% |
| case-14 | ✗→✓ | ▲ Improved | 114% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 45% | 0% |
<!-- Compatibility anchors retained for published 2.x deep links. --> <a id="oceanbase-tenant-management-obd"></a> <a id="when-to-use-this-skill"></a> <a id="tenant-commands"></a> <a id="create-tenant"></a> <a id="drop-tenant"></a> <a id="show-tenants"></a> <a id="optimize-tenant"></a> <a id="backup-restore"></a> <a id="backup--restore"></a> <a id="set-backup-config"></a> <a id="run-backup"></a> <a id="restore-from-backup"></a> <a id="usage-examples"></a> <a id="create-a-tenant"></a> <a id="configure-and-run-backup"></a> <a id="related-skills"></a>
Resolve the installed OBD build, product form, deployment, and tenant before selecting a command. This skill supports community and commercial OceanBase forms when the installed command and component plugins prove the requested operation is available.
Creating a tenant still requires an explicit user request; cluster deployment alone never implies a business tenant. After creation is requested, do not ask for tenant settings the user omitted. Confirm the installed OBD defaults and omit those options so OBD owns the defaults for tenant name, mode, placement, replicas, Units, resources, IOPS, time zone, charset/collation, optimization, variables, network allowlist, and initial password. Explicit user values override only the corresponding defaults.
In the reviewed OBD baseline, the minimal create path produces tenant test, MySQL mode, all active zones, RANDOM primary zone, maximum currently allocatable tenant resources bounded by the selected servers, allowlist %, and an empty tenant password. Preserve that empty-password default when no password was supplied: do not ask for one, generate one, or set one after creation. Read the full tenant default and verification workflow before creating a tenant.
Apply the same no-prompt identity and password policy when building a physical primary/standby tenant pair. If the primary tenant must be created, apply each omitted value independently: the ordinary OBD defaults are tenant test and an empty root password. When creating its standby, omit --tenant-name unless the user supplied a standby name, so OBD uses the primary tenant name; omit --tenant-root-password unless the user supplied the primary tenant's root password, so the reviewed workflow tries the empty-password default. The required positional primary-tenant identity must still be resolved from actual primary-cluster state rather than guessed. These tenant-root defaults do not apply to the separate standbyro replication account, whose password must follow the installed OBD workflow. Read physical primary/standby tenants before creating or operating that relationship.
Read the shared references required by the current stage:
| Request | Required reference | |---|---| | Create or inspect a tenant; plan resources, topology, credentials, or network allowlists | tenant-creation.md | | Drop a tenant | tenant-deletion.md | | Optimize an existing tenant for a workload | workload-optimization.md | | Configure archive/backup storage, run, inspect, or cancel a backup, or restore, inspect, or cancel a tenant restore | backup-restore.md | | Create or operate a physical primary/standby tenant, including switchover, failover, decouple, recovery target, log source, or protection mode | physical-standby.md |
standbyro, storage, decryption, KMS, or other credentials in reusable commands or reports.For cluster deployment, component lifecycle, or upgrades, use cluster-management. For benchmarks and functional tests, use testing-and-benchmark.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→fail | 18,067 | 19,365 | +7% | 1 | 1 | 0% | 2,552 | 3,735 | +46% | 0 | 0 | — |
case-01 | fail→pass | 16,816 | 16,642 | -1% | 1 | 1 | 0% | 1,801 | 3,083 | +71% | 0 | 0 | — |
case-02 | fail→pass | 23,179 | 23,351 | +1% | 1 | 1 | 0% | 2,920 | 4,540 | +55% | 0 | 0 | — |
case-03 | pass→fail | 65,332 | 31,378 | -52% | 1 | 1 | 0% | 8,315 | 4,033 | -51% | 0 | 0 | — |
case-05 | fail→fail | 18,120 | 14,664 | -19% | 1 | 1 | 0% | 2,934 | 3,555 | +21% | 0 | 0 | — |
case-06 | fail→fail | 11,452 | 8,060 | -30% | 1 | 1 | 0% | 1,080 | 2,538 | +135% | 0 | 0 | — |
case-07 | pass→pass | 15,586 | 16,494 | +6% | 1 | 1 | 0% | 1,598 | 3,150 | +97% | 0 | 0 | — |
case-08 | fail→pass | 13,184 | 17,124 | +30% | 1 | 1 | 0% | 2,209 | 3,826 | +73% | 0 | 0 | — |
case-09 | pass→pass | 18,598 | 24,166 | +30% | 1 | 1 | 0% | 2,251 | 4,131 | +84% | 0 | 0 | — |
case-10 | pass→pass | 19,397 | 16,669 | -14% | 1 | 1 | 0% | 2,457 | 3,179 | +29% | 0 | 0 | — |
case-11 | fail→fail | 22,481 | 18,529 | -18% | 1 | 1 | 0% | 2,545 | 3,414 | +34% | 0 | 0 | — |
case-12 | fail→fail | 16,441 | 22,884 | +39% | 1 | 1 | 0% | 1,336 | 4,304 | +222% | 0 | 0 | — |
case-13 | pass→pass | 18,025 | 12,382 | -31% | 1 | 1 | 0% | 1,798 | 2,034 | +13% | 0 | 0 | — |
case-14 | fail→pass | 6,302 | 3,034 | -52% | 1 | 1 | 0% | 768 | 1,643 | +114% | 0 | 0 | — |
case-15 | fail→pass | 14,169 | 5,634 | -60% | 1 | 1 | 0% | 1,387 | 2,009 | +45% | 0 | 0 | — |
case-16 | fail→pass | 13,371 | 10,102 | -24% | 1 | 1 | 0% | 1,951 | 1,969 | +1% | 0 | 0 | — |
case-17 | pass→pass | 20,395 | 12,226 | -40% | 1 | 1 | 0% | 2,074 | 3,151 | +52% | 0 | 0 | — |
case-18 | fail→pass | 22,121 | 14,790 | -33% | 1 | 1 | 0% | 1,200 | 2,622 | +119% | 0 | 0 | — |
case-19 | fail→pass | 10,377 | 9,315 | -10% | 1 | 1 | 0% | 1,375 | 1,749 | +27% | 0 | 0 | — |
case-20 | fail→pass | 14,994 | 11,940 | -20% | 1 | 1 | 0% | 1,455 | 2,215 | +52% | 0 | 0 | — |
case-21 | fail→pass | 17,318 | 19,459 | +12% | 1 | 1 | 0% | 2,653 | 3,027 | +14% | 0 | 0 | — |
case-22 | pass→pass | 9,445 | 13,896 | +47% | 1 | 1 | 0% | 1,369 | 1,542 | +13% | 0 | 0 | — |
case-23 | fail→pass | 16,706 | 3,103 | -81% | 1 | 1 | 0% | 1,736 | 1,683 | -3% | 0 | 0 | — |
case-24 | fail→pass | 8,888 | 14,764 | +66% | 1 | 1 | 0% | 1,184 | 1,705 | +44% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +46 percentage points is the difference between those two pass rates over the 24 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 9/7/2026 | +41% |
| gemini-3.6-flash | verified | 8/6/2026 | +59% |
Other measured skills in the registry, with their headline benchmark lift.