Description
Agent Platform inference is a read-only agent skill for authenticating and connecting to Google Cloud Agent Platform to use generative AI models. It covers three kinds of models. First-party publisher models, which means Gemini, are one section. Third-party publisher models, the OpenMaaS catalog that includes Llama, DeepSeek and Qwen, are another. The third covers custom endpoints, which are any model served on a numeric projects endpoint resource, such as tuned Gemini models, open-source LLMs that you deployed yourself from Model Garden with the deploy skill, and legacy custom models. For each kind, the skill explains how to authenticate and make calls, so the agent can help you get a response from the model you chose. Because it only reads and calls models, it leaves out the safety tiers that apply to changes, and it says so. The skill is aimed at developers who want to add model inference to an application on Google Cloud without working through the platform documentation from scratch. It is published by Google on skills.sh under the google skills repository, with about 21.3K stars and about 7.7K installs. The skills.sh listing shows Gen Agent Trust Hub, Socket and Snyk all passing.…
Highlights
- Authenticates and connects to Agent Platform
- Gemini first-party models
- OpenMaaS third-party models like Llama and Qwen
- Custom endpoints for tuned and self-deployed models
- Read-only: no changes to resources
Ratings & Reviews
No ratings yet
Be the first to rate agent-platform-inference.
People also visit
agent-platform-eval-flywheel
Measures and improves the quality of AI models and agents on Google Cloud
agent-platform-rag-engine-management
Manage and query Agent Platform RAG Engine Corpora
agent-platform-troubleshooting
Troubleshoots Google Cloud Gemini Enterprise Agent Platform issues
agent-platform-inference FAQ
What models does it cover?
First-party Gemini models, third-party OpenMaaS models such as Llama, DeepSeek and Qwen, and custom endpoints.
Can it change my resources?
No. The skill is read-only and does not create or change resources.
What are custom endpoints?
Models served on a numeric endpoint resource, such as tuned Gemini models and open models you deployed from Model Garden.
How do I install it?
Run npx skills add https://github.com/google/skills --skill agent-platform-inference.
What security audits has it passed?
skills.sh lists three audits: Gen Agent Trust Hub Pass, Socket Pass and Snyk Pass.
What are the best agent-platform-inference alternatives?
Popular alternatives to agent-platform-inference include agent-platform-deploy, agent-platform-tuning and agent-platform-eval-flywheel. See all agent-platform-inference alternatives