[!WARNING]
Preview: This feature is subject to the "Pre-GA Offerings Terms" in the
General Service Terms section of the Service Specific
Terms. When you use this
feature with AI Agents, the terms applicable to AI Agents in the Agreement
apply. Pre-GA features are available "as is" and might have limited support.
For more information, see the launch stage
descriptions.
[!NOTE]
To see an example of creating and managing skills in Skill Registry, run the
"Intro to Skill Registry" notebook in one of the following environments:
Open in Colab
Open in Colab Enterprise
Open in Agent Platform Workbench
View on GitHub
Skill Registry
This skill provides instructions for interacting with the Skill Registry on
the Gemini Enterprise Agent Platform. The Skill Registry allows you to create,
update, list, get, delete, and search for skills, as well as manage revisions
and monitor long-running operations.
Core Capabilities
- Skill Discovery — Query the registry to easily search, list, get
specific skills, get skill revisions, and inspect revision histories.
- Skill Lifecycle Management — Create, update, or permanently delete
skills with support for zipped or folder-based payloads.
- Semantic Search — Use the
RetrieveSkills method to find skills by
describing the task you want to accomplish (e.g., "find skills to manage
cloud resources").
- Revision Management — List and inspect individual revisions of a skill.
- Operation Monitoring — Check the completion status of long-running
operations (LROs) returned by create, update, and delete actions.
- Generate Skill — Automate the initial scaffolding of new agent skills
locally.
Core Directives
Prerequisites & Authentication
Permissions
- For read-only access (list, get, search, list revisions, get revisions), the
roles/aiplatform.viewer role is sufficient.
- For write operations (create, update, delete), you need
roles/aiplatform.admin or equivalent.
- Skill Registry inherits project-level permissions.
Library & Authentication
# Install required libraries
pip install google-auth requests
# Authenticate with Google Cloud
gcloud auth application-default login
Environment Variables
The following variables are required for operations:
GCP_PROJECT_ID: Your Google Cloud project ID.
GCP_LOCATION: The region (e.g., us-central1). See Available
regions below.
Compliance
| Feature |
Status |
| Access Transparency |
Supported |
| Data Residency (DRZ) in US and EU |
Supported |
| Customer-Managed Encryption Keys (CMEK) |
Not Supported |
| HIPAA certification |
Not Supported |
| VPC-SC |
Not Supported |
Available regions
| Region |
Location |
us-central1 |
Iowa |
europe-west4 |
Netherlands |
us-east5 |
Columbus, Ohio |
Skill ID constraints
The SKILL_ID is immutable and remains permanently reserved once created,
even if you delete the skill later. Skill IDs must adhere to the following
constraints:
- Must be 1 to 63 characters long.
- Must contain only lowercase letters, numbers, and hyphens.
- Must start with a letter and end with a letter or number.
- Must not start with
gcp- (this prefix is reserved for built-in skills).
Note: Use a descriptive SKILL_ID. Since the SKILL_ID becomes the
folder name when the skill is attached to an agent, a meaningful name helps
the model understand the skill's purpose.
Caution: Built-in skills with IDs starting with gcp- cannot be deleted.
When you delete a skill, its SKILL_ID is reserved and cannot be reused for
24 hours.
Quickstart
Quickly search for available skills in the registry:
python3 scripts/skill_registry_ops.py search \
--query "test skill" \
--top-k 5
Operations
- Skill Discovery: query-skills.md — Search,
list, get, retrieve skills, and inspect revision histories with REST,
Python, and Node.js examples.
- Skill Lifecycle: manage-skills.md —
Create, update, and delete skills with REST, Python, and Node.js examples.
- Monitor Operations:
monitor-operations.md — Check the
status of long-running operations.
- Generate Skill: generate-skill.md —
Scaffold new agent skills locally.
What's next
- Attach skills to an agent: Learn how to attach registered skills when
creating or updating an agent using the Managed Agents API on Agent Platform.
See guide
- ADK Skill Registry integration: Learn how to integrate with Skill
Registry in the Agent Development Kit (ADK).
See documentation
1---2name: agent-platform-skill-registry3description: Interact with the Gemini Enterprise Agent Platform Skill Registry to create, update, manage, and search for skills. Use this skill to enable agents to register new functionality, discover existing capabilities, manage skill revisions, and monitor long-running operations.4---5
6> [!WARNING]
7> **Preview:** This feature is subject to the "Pre-GA Offerings Terms" in the
8> General Service Terms section of the [Service Specific
9> Terms](https://cloud.google.com/terms/service-terms#1). When you use this
10> feature with AI Agents, the terms applicable to AI Agents in the Agreement
11> apply. Pre-GA features are available "as is" and might have limited support.
12> For more information, see the [launch stage
13> descriptions](https://cloud.google.com/products#product-launch-stages).
14
15> [!NOTE]
16> To see an example of creating and managing skills in Skill Registry, run the
17> "Intro to Skill Registry" notebook in one of the following environments:
18>
19> [Open in Colab](https://colab.research.google.com/github/GoogleCloudPlatform/generative-ai/blob/main/agents/skill-registry/intro_skill_registry.ipynb)
20>
21> [Open in Colab Enterprise](https://console.cloud.google.com/agent-platform/colab/import/https%3A%2F%2Fraw.githubusercontent.com%2FGoogleCloudPlatform%2Fgenerative-ai%2Fmain%2Fagents%2Fskill-registry%2Fintro_skill_registry.ipynb)
22>
23> [Open in Agent Platform Workbench](https://console.cloud.google.com/agent-platform/workbench/deploy-notebook?download_url=https://raw.githubusercontent.com/GoogleCloudPlatform/generative-ai/main/agents/skill-registry/intro_skill_registry.ipynb)
24>
25> [View on GitHub](https://github.com/GoogleCloudPlatform/generative-ai/blob/main/agents/skill-registry/intro_skill_registry.ipynb)
26
27# Skill Registry
28
29This skill provides instructions for interacting with the **Skill Registry** on
30the Gemini Enterprise Agent Platform. The Skill Registry allows you to create,
31update, list, get, delete, and search for skills, as well as manage revisions
32and monitor long-running operations.
33
34## Core Capabilities
35
36- **Skill Discovery** — Query the registry to easily search, list, get
37 specific skills, get skill revisions, and inspect revision histories.
38- **Skill Lifecycle Management** — Create, update, or permanently delete
39 skills with support for zipped or folder-based payloads.
40- **Semantic Search** — Use the `RetrieveSkills` method to find skills by
41 describing the task you want to accomplish (e.g., "find skills to manage
42 cloud resources").
43- **Revision Management** — List and inspect individual revisions of a skill.
44- **Operation Monitoring** — Check the completion status of long-running
45 operations (LROs) returned by create, update, and delete actions.
46- **Generate Skill** — Automate the initial scaffolding of new agent skills
47 locally.
48
49## Core Directives
50
51- **Mandatory Validation**: ALWAYS execute the environment validation check
52 before performing any operations.
53
54 ```bash
55 python3 scripts/validate_env.py
56 ```
57
58## Prerequisites & Authentication
59
60### Permissions
61
62- For read-only access (list, get, search, list revisions, get revisions), the
63 `roles/aiplatform.viewer` role is sufficient.
64- For write operations (create, update, delete), you need
65 `roles/aiplatform.admin` or equivalent.
66- Skill Registry inherits project-level permissions.
67
68### Library & Authentication
69
70```bash
71# Install required libraries
72pip install google-auth requests
73
74# Authenticate with Google Cloud
75gcloud auth application-default login
76```
77
78### Environment Variables
79
80The following variables are required for operations:
81
82- `GCP_PROJECT_ID`: Your Google Cloud project ID.
83- `GCP_LOCATION`: The region (e.g., `us-central1`). See [Available
84 regions](#available-regions) below.
85
86## Compliance
87
88| Feature | Status |
89|---|---|
90| Access Transparency | Supported |
91| Data Residency (DRZ) in US and EU | Supported |
92| Customer-Managed Encryption Keys (CMEK) | Not Supported |
93| HIPAA certification | Not Supported |
94| VPC-SC | Not Supported |
95
96## Available regions
97
98| Region | Location |
99|---|---|
100| `us-central1` | Iowa |
101| `europe-west4` | Netherlands |
102| `us-east5` | Columbus, Ohio |
103
104## Skill ID constraints
105
106The `SKILL_ID` is **immutable** and remains permanently reserved once created,
107even if you delete the skill later. Skill IDs must adhere to the following
108constraints:
109
110- Must be 1 to 63 characters long.
111- Must contain only lowercase letters, numbers, and hyphens.
112- Must start with a letter and end with a letter or number.
113- Must not start with `gcp-` (this prefix is reserved for built-in skills).
114
115> **Note:** Use a descriptive `SKILL_ID`. Since the `SKILL_ID` becomes the
116> folder name when the skill is attached to an agent, a meaningful name helps
117> the model understand the skill's purpose.
118
119> **Caution:** Built-in skills with IDs starting with `gcp-` cannot be deleted.
120> When you delete a skill, its `SKILL_ID` is reserved and cannot be reused for
121> 24 hours.
122
123## Quickstart
124
125Quickly search for available skills in the registry:
126
127```bash
128python3 scripts/skill_registry_ops.py search \
129 --query "test skill" \
130 --top-k 5
131```
132
133---
134
135## Operations
136
137- **Skill Discovery**: [query-skills.md](references/query-skills.md) — Search,
138 list, get, retrieve skills, and inspect revision histories with REST,
139 Python, and Node.js examples.
140- **Skill Lifecycle**: [manage-skills.md](references/manage-skills.md) —
141 Create, update, and delete skills with REST, Python, and Node.js examples.
142- **Monitor Operations**:
143 [monitor-operations.md](references/monitor-operations.md) — Check the
144 status of long-running operations.
145- **Generate Skill**: [generate-skill.md](references/generate-skill.md) —
146 Scaffold new agent skills locally.
147
148## What's next
149
150- **Attach skills to an agent**: Learn how to attach registered skills when
151 creating or updating an agent using the Managed Agents API on Agent Platform.
152 [See guide](https://docs.cloud.google.com/gemini-enterprise-agent-platform/build/managed-agents/create-manage#create-attach-skills)
153- **ADK Skill Registry integration**: Learn how to integrate with Skill
154 Registry in the Agent Development Kit (ADK).
155 [See documentation](https://adk.dev/integrations/skills-registry/)