gemini-agents-api Skill (Google) for Claude: What It Does and How to Install It

Makes Claude create, list, update and delete custom agents on Google's Gemini Enterprise Agent Platform.

Installs: 9.9K; on skills.sh
Stars: 21.8K; on GitHub
Files: 1; besides SKILL.md
Updated: Nov 2026; last repo commit

This skill gives Claude the REST requests for the Managed Agents API on Google's Gemini Enterprise Agent Platform. It covers creating an agent, checking the long-running operation, getting, listing, updating and deleting agents, and attaching tools and MCP servers.

It needs the gcloud CLI and a Google Cloud project, and Claude runs gcloud and curl against your project with your token. It is in Google's own google/skills repo. It bundles no scripts. Some calls change or delete agents, so approve them one by one.

Good for

Developers building stateful agents on Google Cloud
Platform teams scripting agent setup with files, skills and tools mounted
Teams adding MCP servers to Google-hosted agents

Install gemini-agents-api

Claude Code

Copy this prompt.

Install the gemini-agents-api skill in Claude Code. Run this command:

npx skills add https://github.com/google/skills --skill gemini-agents-api

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js, gcloud CLI.

The skill drives gcloud and curl, so it needs the gcloud CLI and a Google Cloud project. Because it runs local commands, there is no Claude app upload route. The repo README also supports choosing skills and target agents interactively with npx skills add google/skills.

Built from the package data we hold. Not yet tested by us.

What's in the skill

When it loads

Manages custom Agent resources on Gemini Enterprise Agent Platform. Use when the user wants to programmatically create, configure, list, update, or delete stateful, server-managed Agent resources (including mounting files, skills, and tools) before executing conversations.

Sections in SKILL.md

Attribution
1. Authentication & Setup
1. Setup Environment Variables
2. Endpoint URL
2. Programmatic Agent Management (Control Plane CRUD)
1. Create Agent (Long-Running Operation)
2. Polling the LRO Status
3. Get Agent
4. List Agents
5. Update Agent (Patching Configuration)
6. Delete Agent
3. Interacting with Custom Agents (Data Plane)

Key facts

Author: Google (the vendor's own repo, google/skills)
License: Apache-2.0
Installs on skills.sh: 9,909 (Nov 2, 2026)
GitHub stars: 21,764 for the whole google/skills repo (Nov 2, 2026)
Last repo commit: Nov 1, 2026
Audits on skills.sh: Gen Agent Trust Hub: pass, Socket: pass, Snyk: pass
What it includes: 2 files including SKILL.md: references/attribution.md (no scripts)
Needs: gcloud CLI and a Google Cloud project with the Managed Agents API available
Skill version: 1.0.1 (SKILL.md metadata)

What it includes

Tags gcloud and HTTP calls with a skill name and version
references/attribution.md says the tag carries no user data. You can ask to turn it off.
Read · Attribution rules
Sets PROJECT_ID, LOCATION and an ACCESS_TOKEN from gcloud
The token comes from gcloud auth print-access-token and goes in a Bearer header.
Runs code · 1. Authentication and setup
POST to the agents endpoint on aiplatform.googleapis.com
A long-running operation. Can mount files or skills from Cloud Storage or the Skill Registry and attach tools such as code execution, filesystem, google_search and url_context.
Runs code · 2. Create agent
GET the operation until done is true
Returns the finished Agent resource.
Runs code · Poll the operation
GET one agent's configuration
Shows its tools and environment setup.
Runs code · 3. Get agent
GET all custom agents in a project
Uses the project and location you set.
Runs code · 4. List agents
PATCH with a required update_mask
SKILL.md tells Claude to always explain that update_mask names the fields being changed.
Runs code · 5. Update agent
DELETE an agent resource
Frees the agent's backend workspace container.
Delete · 6. Delete agent
Puts an MCP server URL and optional headers under tools
Headers can carry a bearer token, which the platform sends only to that URL, per SKILL.md.
Write · Add MCP servers to an agent
Points to the gemini-interactions-api skill
Chatting with a created agent is a separate API covered by that skill.
Read · Hand-off to conversations

From the official docs

Permissions

High risk

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

It can create, update and delete agents
SKILL.md documents POST, PATCH and DELETE calls to aiplatform.googleapis.com with your bearer token. Approve each call and check the project, location and agent id. github.com
The example agent allows every domain
The create example sets the agent's network allowlist to the domain "*". Narrow it before you use a config like that for real. github.com
MCP headers can hold secrets
The MCP example puts an Authorization bearer token in the agent's tool config. Keep real tokens out of chat and out of shared config. github.com
Calls carry a usage tag
gcloud calls get a CLOUDSDK_METRICS_REQUEST_ATTRIBUTION value and HTTP calls a User-Agent suffix naming the skill. attribution.md says it carries no user data and can be disabled on request. github.com

Security advisories

No published security advisories as of Nov 2, 2026.

Technical details

Turning on read-only	No read-only mode. Ask Claude to only list or get agents, or use a token with limited roles.
Talks to	aiplatform.googleapis.com (Agent Platform REST API). Cloud Storage (only if you mount a bucket)
Sign-in and keys	SKILL.md says requests need a Bearer token derived from Application Default Credentials, taken with gcloud auth print-access-token into an ACCESS_TOKEN variable. It does not say where credentials are stored.
Scopes	Uses a Google Cloud access token from gcloud auth print-access-token Creates, updates and deletes Agent resources in a project Can mount Cloud Storage buckets or Skill Registry skills into an agent

Questions

What does the gemini-agents-api skill do?

It makes Claude manage custom agent resources on Google's Gemini Enterprise Agent Platform through REST calls. Claude can create, get, list, update and delete agents, and mount files, skills and tools into them.

Is this made by Google?

It is in Google's own google/skills repo, which its README describes as Agent Skills for Google products. The repo owner is Google.

How do I install it?

Use the install box at the top of this page. The repo README documents the skills CLI with google/skills, which lets you pick skills and target agents. The skill runs gcloud and curl locally, so we do not offer a Claude app upload route.

What do I need before using it?

A Google Cloud project, the gcloud CLI signed in with credentials, and access to the Managed Agents API. SKILL.md sets PROJECT_ID, LOCATION and an access token from gcloud.

Can it delete things?

Yes. It documents a DELETE request for an agent resource and an update request that changes configuration in place. Check the agent id before you approve either one.

Does it send usage data to Google?

It adds a tag to gcloud commands and to the User-Agent header on direct HTTP calls. references/attribution.md says the tag identifies only the skill and its version, and that you can ask to disable it.

How do I chat with the agent I created?

SKILL.md says that is the Interactions API and refers you to the gemini-interactions-api skill in the same repo. Install that skill too if you want Claude to help with conversations.

Is it safe?

It bundles no scripts, but it has Claude run gcloud and curl calls against your Google Cloud project with your token. Some can update or delete agents, so approve each call and use a project you can afford to change.

Information

updated Nov 2, 2026

Repository: google/skills
Folder: skills/cloud/gemini-agents-api

google/skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 3

gemini-agents-api SKILL.md · github.com
attribution.md · github.com
google/skills README · github.com