Zilliz Cloud CLI Skill

Your Zilliz Cloud console in the terminal. This skill enables you to operate the zilliz-cli, bringing the full power of Zilliz Cloud — cluster management, collection operations, vector search, RBAC, backups, and more — through natural language.

When to Use

Use this skill when the user wants to:

• - Manage Zilliz Cloud clusters (create, delete, suspend, resume, modify)
• Work with Milvus collections (create, describe, load, release, drop)
• Perform vector operations (search, query, insert, upsert, delete)
• Manage indexes, databases, partitions
• Set up users, roles, and access control (RBAC)
• Create and manage backups or backup policies
• Import bulk data from cloud storage
• Check billing, usage, or invoices
• Monitor cluster and collection status

Capabilities Overview

Area	What You Can Do
Clusters	Create, delete, suspend, resume, modify
Collections

Create with custom schema, load, release, rename, drop | | Vectors | Search, query, insert, upsert, delete | | Indexes | Create, list, describe, drop | | Databases | Create, list, describe, drop | | Users & Roles | RBAC setup, privilege management | | Backups | Create, restore, export, policy management | | Import | Bulk data import from cloud storage | | Partitions | Create, load, release, manage | | Monitoring | Cluster status, collection stats | | Billing | Usage, invoices, payment methods | | Projects | Project and region management |

Requirements

• - Python 3.10+
• A Zilliz Cloud account (or local Milvus instance)

---

Setup & Authentication

Prerequisites

Before running any zilliz-cli command, verify the following in order:

1. 1. CLI installed and up to date? Run python3 -m pip install --upgrade zilliz-cli to ensure the latest version is installed.
2. Logged in? Run zilliz auth status. If not logged in, guide through login (see below).
3. Context set? (Only for data-plane operations) Run zilliz context current. If no context, guide through context setup.

Install / Upgrade CLI

CODEBLOCK0

Verify installation:

CODEBLOCK1

Authentication

IMPORTANT: Login commands (zilliz login, zilliz configure) require an interactive terminal and CANNOT run inside the AI agent. Always instruct the user to run these in their own terminal.

Check if already logged in:

CODEBLOCK2

If not logged in, tell the user to open their own terminal and run one of the following:

Option 1: Browser-based login (OAuth) — full feature access

CODEBLOCK3

• - Opens a browser for authentication
• Retrieves user info, organization data, and API keys
• Use --no-browser in headless environments (displays a URL to visit manually)

Option 2a: API Key via login command

CODEBLOCK4

Option 2b: API Key via configure (legacy)

CODEBLOCK5

• - Prompts for an API key (found in Zilliz Cloud console under API Keys)
• Limitations compared to OAuth login:

- Organization switching not available - On Serverless clusters: database management, user/role management may be restricted - Some control-plane operations may require OAuth login

Option 3: Environment variable

User can add to their shell profile (.zshrc / .bashrc):

CODEBLOCK6

After the user completes authentication, verify by running:

CODEBLOCK7

Configure Subcommands

CODEBLOCK8

Switch Organization

These commands require an interactive terminal. Instruct the user to run in their own terminal:

CODEBLOCK9

Logout

CODEBLOCK10

Set Cluster Context

Data-plane commands (collection, vector, index, etc.) require an active cluster context.

CODEBLOCK11

View Current Context

CODEBLOCK12

Output Format

All zilliz-cli commands support --output json for structured, machine-readable output:

CODEBLOCK13

Available formats: json, table, text. Default is text.

Cluster Type Differences

Feature	Free	Serverless	Dedicated
Collection CRUD	Yes	Yes	Yes
Vector search/query

Yes | Yes | Yes | | Database create/drop | No | No | Yes | | User/role management | No | Limited | Yes | | Backup management | No | Yes | Yes | | Cluster modify | No | No | Yes |

---

Quickstart Command

Guide the user through the complete Zilliz Cloud CLI setup:

1. 1. Check Python — python3 --version (need 3.10+)
2. Install CLI — python3 -m pip install --upgrade zilliz-cli then INLINECODE16
3. Authenticate — zilliz auth status; if not logged in, instruct user to run zilliz login in their own terminal
4. List clusters — zilliz cluster list; if none, offer to create one
5. Set context — INLINECODE20
6. Verify — zilliz context current and INLINECODE22

---

Cluster Management

List Clusters

CODEBLOCK14

Describe a Cluster

CODEBLOCK15

Create a Cluster

CODEBLOCK16

To find available project IDs, cloud providers, and regions:

CODEBLOCK17

Delete / Suspend / Resume a Cluster

CODEBLOCK18

Modify a Cluster

CODEBLOCK19

Cluster Guidance

• - Before creating a cluster, help the user choose a region by running zilliz cluster providers and zilliz cluster regions.
• Cluster creation is asynchronous. After cluster create, poll with zilliz cluster describe until status becomes RUNNING.
• Before deleting a cluster, always confirm with the user — this is irreversible.
• After creating a cluster, suggest setting context with zilliz context set --cluster-id <id>.

---

Collection Management

All collection commands accept an optional --database <db-name> flag.

List / Describe Collections

CODEBLOCK20

Create a Collection

Quick create with auto schema:

CODEBLOCK21

Advanced create with custom schema:

CODEBLOCK22

Drop / Rename / Check Existence

CODEBLOCK23

Load / Release / Stats

CODEBLOCK24

Collection Aliases

CODEBLOCK25

Collection Guidance

• - Ask about use case to recommend dimension, metric type, and schema.
• Before dropping a collection, confirm with the user — this deletes all data.
• A collection must be loaded before search or query.
• Use describe to inspect schema before vector operations.

---

Vector Operations

All vector commands accept an optional --database <db-name> flag. Target collection must exist and be loaded.

Vector Search

CODEBLOCK26

Hybrid Search

CODEBLOCK27

Query (Filter-Based Retrieval)

CODEBLOCK28

Get by ID

CODEBLOCK29

Insert / Upsert Vectors

CODEBLOCK30

For large datasets, read from file:

CODEBLOCK31

Delete Vectors

CODEBLOCK32

Filter Expression Syntax

Expression	Example
Comparison	INLINECODE32
Equality

status == "active" | | IN list | id in [1, 2, 3] | | AND/OR | age > 20 and status == "active" | | String match | text like "hello%" | | Array contains | tags array_contains "ml" |

Vector Guidance

• - Before searching, inspect schema with zilliz collection describe for field names and dimensions.
• The --data value must match the collection's vector dimension exactly.
• For text-to-vector search, the user needs an embedding model to convert text to vectors first.
• For large insert operations, suggest writing data to a JSON file first.

---

Index Management

All index commands accept an optional --database <db-name> flag.

Create an Index

CODEBLOCK33

Common index types: AUTOINDEX (recommended), IVF_FLAT, IVF_SQ8, HNSW.
Common metric types: COSINE, L2, IP.

List / Describe / Drop Indexes

CODEBLOCK34

Index Guidance

• - On Zilliz Cloud, AUTOINDEX is recommended for most use cases.
• An index is required before loading a collection for search.
• After creating an index, remind the user to load the collection.

---

Database Management

Commands

CODEBLOCK35

Database Guidance

• - Database create, describe, and drop are only available on Dedicated clusters. database list works on all cluster types.
• Every cluster has a "default" database.
• Before dropping a database, confirm with the user — all collections in it will be deleted.
• After creating a database, suggest switching context: zilliz context set --database <db-name>.

---

Partition Management

Commands

CODEBLOCK36

Partition Guidance

• - Every collection has a default _default partition.
• A partition must be loaded before it can be searched.
• Before dropping a partition, confirm with the user.

---

User & Role Management (RBAC)

Only available on Dedicated clusters.

Users

CODEBLOCK37

Roles

CODEBLOCK38

Common privileges:

• - Collection: Search, Query, Insert, Delete, CreateIndex, INLINECODE57
• Global: CreateCollection, INLINECODE59
• Database: INLINECODE60

RBAC Guidance

• - Built-in roles: admin (full access), public (no privileges by default).
• Suggested workflow: create role → grant privileges → create user → assign role.
• Use * as object-name to grant privilege on all objects of that type.

---

Backup Management

No cluster context required — backup operations use --cluster-id directly.

Create a Backup

CODEBLOCK39

List / Describe / Delete Backups

CODEBLOCK40

Export a Backup

CODEBLOCK41

Restore to a New Cluster

CODEBLOCK42

Restore Specific Collections

CODEBLOCK43

Backup Policy

CODEBLOCK44

Backup Guidance

• - Before deleting a backup, confirm with the user — this is irreversible.
• Explain difference: cluster restore = new cluster, collection restore = into existing cluster.
• Suggest backup policies for production clusters.

---

Data Import

Target collection must exist on the target cluster.

Commands

CODEBLOCK45

Supported file formats: Parquet, JSON, CSV.

Import Guidance

• - Import jobs run asynchronously — use import status to track progress.
• Data files must be accessible from Zilliz Cloud (e.g., S3, GCS with integration configured).
• Collection schema must match the data file structure.

---

Billing

Requires OAuth login (not API Key mode).

Commands

CODEBLOCK46

Billing Guidance

• - The bind-card command handles sensitive card data — always instruct the user to run it in their own terminal.

---

Project & Region Management

Projects

CODEBLOCK47

Volumes

CODEBLOCK48

---

Monitoring & Status

Status Overview Workflow

When the user asks for status, gather and present:

1. 1. INLINECODE67
2. INLINECODE68
3. INLINECODE69
4. INLINECODE70
5. For each collection: zilliz collection get-stats --name <name> --output json and INLINECODE72

Present as:

Cluster: \ (\)
Status: RUNNING | Region: \ | Plan: \
Database: \

Collections:

Collection	Rows	Load State
my_collection	10,000	Loaded

---

General Guidance

• - Always check prerequisites (CLI installed, logged in, context set) before executing commands.
• NEVER run zilliz login, zilliz configure, or zilliz auth switch inside the AI agent — they require interactive input.
• NEVER ask the user to paste API keys into the chat — guide them to configure credentials in their own terminal.
• Before any destructive operation (delete cluster, drop collection, drop database, delete backup), always confirm with the user.
• Use --output json when parsing results programmatically, then format into readable summaries.
• When a command fails with permissions error, check the cluster type first.
• When a command fails unexpectedly, consider whether the cluster type or auth mode may be the cause.