zilliz
Manage Zilliz Cloud vector databases via zilliz-cli. Use when the user wants to: (1) set up zilliz-cli (install, login, configure), (2) manage clusters (create, list, describe, suspend, resume, delete, modify), (3) manage collections, indexes, partitions, databases, or aliases, (4) perform vector operations (search, query, insert, upsert, delete, hybrid search), (5) manage database users, roles, and RBAC privileges, or cloud-level access control (org/project roles, member and group grants), (6) handle backups, imports, volumes, storage integrations, billing, monitoring, projects, regions, or async jobs. Triggers on: zilliz, vector database, semantic search, RAG, embeddings, collection schema, vector index, similarity search, zilliz-cli.
How do I install this agent skill?
npx skills add https://github.com/zilliztech/zilliz-skill --skill zillizIs this agent skill safe to install?
- Gen Agent Trust Hubpass
This skill is a secure and well-structured tool for managing Zilliz Cloud and Milvus databases. It uses the vendor's official CLI and installation scripts, and it includes robust safety instructions to prevent the exposure of sensitive credentials to the agent.
- Socketpass
No alerts
- Snykpass
Risk: LOW · No issues
- ZeroLeakspass
1 finding · Score: 82/100
What does this agent skill do?
Zilliz Cloud CLI Skill
Operate Zilliz Cloud through zilliz-cli. All operations are executed via shell commands.
Prerequisite Check (always run first)
Before any operation, verify in order:
# 1. Install or update the CLI
curl -fsSL https://raw.githubusercontent.com/zilliztech/zilliz-cli/master/install.sh | bash
# 2. Logged in? If not: guide user to login in their terminal
zilliz auth status
# 3. Context set? (only for data-plane ops)
zilliz context current
CRITICAL: zilliz login, zilliz configure, and zilliz auth switch require interactive input — NEVER run them in a non-interactive shell. Instruct the user to run in their own terminal. NEVER ask for API keys in chat.
Command Pattern
zilliz <resource> <action> --flag <value> [--optional-flag <value>]
All commands support --output json|table|text (default: text). Use --output json for programmatic parsing.
Cluster Type Capabilities
| Feature | Free | Serverless | Dedicated |
|---|---|---|---|
| Collection CRUD & Vector ops | Yes | Yes | Yes |
| Database create/drop | No | No | Yes |
| User/role management | No | Limited | Yes |
| Backup management | No | Yes | Yes |
| Cluster modify (CU/replica) | No | No | Yes |
Check cluster type first when a command fails with permission errors.
Reference Files
Each reference covers one resource domain with full command syntax, options, and guidance. Read the relevant reference when handling that domain:
| Domain | Reference | When to read |
|---|---|---|
| Setup & auth | references/setup.md | Install, login, context, config, troubleshooting |
| Clusters | references/cluster.md | Create, list, describe, modify, suspend, resume, delete clusters; create Vector Lake |
| On-demand clusters | references/on-demand-cluster.md | Create, list, describe, delete on-demand (Vector Lake) clusters with auto-suspend TTL |
| Collections | references/collection.md | Create, list, describe, drop, rename, load, release, aliases, per-collection metrics |
| External collections | references/external-collection.md | Trigger, describe, list refresh jobs for Vector Lake-backed collections |
| Vectors | references/vector.md | Search, query, insert, upsert, get, delete, hybrid search, filters |
| Indexes | references/index.md | Create, list, describe, drop indexes |
| Databases | references/database.md | Create, list, describe, drop databases |
| Partitions | references/partition.md | Create, list, load, release, drop partitions |
| Users & roles | references/user-role.md | RBAC: users, roles, privileges (Dedicated only) |
| Access control | references/acl.md | Cloud-level RBAC: custom org/project roles, policies, member and SCIM group grants |
| Backups | references/backup.md | Create, restore, export backups; manage backup policies |
| Import | references/import.md | Bulk data import from object storage or files uploaded to a volume |
| Billing | references/billing.md | Usage queries, invoices, payment methods |
| Monitoring | references/monitoring.md | Cluster status, collection stats, load states, cluster and per-collection time-series metrics |
| Projects & regions | references/project-region.md | Create, list, describe, delete, upgrade projects; bind regions |
| Volumes | references/volume.md | Create, list, describe, delete volumes; upload local files for import |
| Storage integrations | references/storage-integration.md | Connect your own S3 / Azure Blob / GCS bucket: auth policy, validate, manage |
| PrivateLink | references/privatelink.md | Endpoint services, project endpoints, whitelist management |
| Jobs | references/job.md | Track async job status, wait for completion |
| Diagnose | references/diagnose.md | Read-only health diagnosis for clusters and collections (slow, stuck, errors, quota) |
Quick-Start Workflow
For new users, guide through setup in order:
- Install CLI:
curl -fsSL https://raw.githubusercontent.com/zilliztech/zilliz-cli/master/install.sh | bash - Authenticate: user runs
zilliz loginin their terminal - Verify:
zilliz auth status - List clusters:
zilliz cluster list - Set context:
zilliz context set --cluster-id <id> - Verify:
zilliz context current && zilliz collection list
Common Workflows
Create collection and prepare for search:
zilliz collection create --name <name> --dimension <dim>— createzilliz index create --collection <name>— index (AUTOINDEX recommended)zilliz collection load --name <name>— load into memoryzilliz vector search --collection <name> --data '[[...]]'— search
Status overview (read references/monitoring.md):
zilliz context current --output jsonzilliz cluster describe --cluster-id <id> --output jsonzilliz database list --output json- For each DB:
collection list,get-stats,get-load-state,index list - Present as formatted table
Async Operations
These return immediately with a jobId; track with job describe (see references/job.md):
| Operation | Poll command |
|---|---|
cluster create | cluster describe --cluster-id <id> until RUNNING |
backup create/export/restore-* | job describe --job-id <jid> or backup describe |
import start | job describe --job-id <jid> |
| Any async op with jobId | job describe --job-id <jid> --wait |
Safety Rules
- Confirm before destructive ops: drop collection/database, delete cluster/backup/volume/storage integration, drop user/role
- Sensitive commands in user's terminal:
zilliz login,zilliz configure,zilliz billing bind-card - Never expose credentials in chat or command output
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/zilliztech/zilliz-skill/zilliz">View zilliz on skillZs</a>