--- title: "Agent Platform GenAI Inference: review, setup & limitations · Undominated.ai" canonical: https://undominated.ai/skills/google-agent-platform-inference/ description: "Calls Gemini and third-party OpenMaaS models on Google Cloud Agent Platform, asking for model, project and region instead of defaulting." --- # Agent Platform GenAI Inference: review, setup & limitations · Undominated.ai > Calls Gemini and third-party OpenMaaS models on Google Cloud Agent Platform, asking for model, project and region instead of defaulting. [← Explore all skills](/skills/) AI APPLICATION DEVELOPMENT / Google # Agent Platform GenAI Inference Calls Gemini and third-party OpenMaaS models on Google Cloud Agent Platform, asking for model, project and region instead of defaulting. [See setup guidance ↓](#setup)[Original source ↗](https://github.com/google/skills/tree/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference) SOURCE REVIEW Reviewed 2026-10-13 Evidence 4 linked sources Publisher Google Licence [Apache-2.0 ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/LICENSE) Revision 4601752b28d6 [Read what was—and wasn’t—checked ↓](#review) “Connects to and performs inference with Google Cloud Agent Platform GenAI models, including First-Party Gemini models and Third-Party OpenMaaS models (Llama, DeepSeek, Qwen, etc.). Use when asked to perform inference, ask a model a question, run a test prompt, execute chat completions, or generate code for calling Gemini or OpenMaaS models, authenticate with GenAI SDK, OpenAI SDK, or legacy Agent Platform SDK, configure base URLs and global/regional endpoints, or troubleshoot 429 Resource Exhausted (DSQ), 400 User Validation, or 404 Not Found errors. Don't use for deploying models to endpoints or for running model evaluations.” [Google · upstream description ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference/SKILL.md) Our analysis follows below. 01 / THE REASONING ## Why this made the selection. - The file says never to guess, volunteer or default to a model, and to ask for the project ID and region rather than assume them. - Every inference needs a confirmation card listing project, region, model ID, SDK and input, and the reply must restate those parameters with the model output. ### A good fit for - Calling Gemini or an OpenMaaS model such as Llama, DeepSeek or Qwen on Agent Platform - Diagnosing 429, 400 or 404 errors from Agent Platform inference - Locate an active deployed endpoint and select its shared or dedicated request host. ### Weigh up before choosing - Region availability is checked with a live generateContent call to the exact model and region, which the file requires before answering region-specific requests. - Model IDs in the examples are illustrative and the file points to the linked model pages. The bundled scripts were not opened, and the README install command does not pin this revision. - The region-probe section says to stop on a target-region 404 without silently changing regions, while the later troubleshooting section suggests switching to us-central1 or global; resolve that inconsistency against the requested region before use. 02 / THE REVIEW RECORD ## What we actually inspected. Source review has boundaries. A clear record is more useful than a “safe” badge. ### Material inspected - skills/cloud/agent-platform-inference/SKILL.md (frontmatter, safety and confirmation tiers, missing-parameter rules, region availability check, authentication lines) - README.md (installation, skill list, licence) - LICENSE Apache-2.0 text - skills/cloud/agent-platform-inference/SKILL.md - README.md - LICENSE ### Our findings - The confirmation card must list project, region, model ID, SDK and input; a paraphrase is not sufficient, per the file. - Google Cloud access tokens used as OpenAI SDK keys typically expire after an hour, per the file. - The endpoint section distinguishes active serving endpoints from historical tuning-job endpoint identifiers and directs the caller to inspect dedicatedEndpointDns when choosing the request host. ### Not established by this review - The installer was not run, and the skill was not executed against a project or service. No inference or region probe was run. - The scripts/ directory and the linked model documentation were not opened. - Not executed: it was not installed or run. The review applies to the material and revision named here. A newer upstream release can change its behavior. 03 / PUT IT TO WORK ## Add a skill to your workflow. [Upstream setup instructions ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/README.md) DOCUMENTED COMMAND npx skills add google/skills Copy command ↗ Run in your terminal to choose your AI tools, project or global scope, and installation method. This short command follows the current upstream version; the review covers the revision shown above. - The README documents npx skills add google/skills and says to select specific skills, target agents and scope when prompted. Select agent-platform-inference. - The command does not pin revision 4601752b28d601d54efb39b06a37488107fc6b1f. Compare the installed directory with that revision. - The file's authentication section uses gcloud auth login and gcloud auth application-default login; the OpenAI SDK path uses a Google Cloud access token as its API key. ### Before you start - Node.js and npm for the documented Skills installer - The gcloud CLI with Application Default Credentials and a project with Agent Platform access ### Compatibility Agent Skills-compatible hosts · Claude Code · Codex · Antigravity CLI · Google Cloud Agent Platform ### Implementation Python ### Google Cloud project inference calls - Runs inference with the Google GenAI SDK, the OpenAI SDK or the legacy Agent Platform SDK using the user's Google Cloud credentials. - Each inference requires interactive confirmation; region probes call the exact model the user named. ### Cost model Source is Apache-2.0. Inference and probe calls are billed to the user's Google Cloud project. Those charges were not measured. 04 / FOLLOW THE EVIDENCE ## The source trail. Our notes are separate from the original resource. Check upstream before adopting a new version. - [Skill definition ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference/SKILL.md) Checked 2026-10-13 https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference/SKILL.md Supports: summary, upstreamDescription, whySelected, limitations, access, review - [Install instructions and skill index ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/README.md) Checked 2026-10-13 https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/README.md Supports: install, compatibility - [Repository licence ↗](https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/LICENSE) Checked 2026-10-13 https://github.com/google/skills/blob/4601752b28d601d54efb39b06a37488107fc6b1f/LICENSE Supports: license - [Pinned continuation review W39 ↗](https://raw.githubusercontent.com/google/skills/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference/SKILL.md) Checked 2026-10-15 https://raw.githubusercontent.com/google/skills/4601752b28d601d54efb39b06a37488107fc6b1f/skills/cloud/agent-platform-inference/SKILL.md Supports: review.findings, bestFor, limitations KEEP COMPARING ## Other approaches to consider. Related by category or shared topics. These are alternatives to inspect, not a measured quality order. [### Neon AI Gateway ↗ Adds multi-provider LLM calling through Neon's branch-scoped gateway with one credential, resolving model IDs against the live catalog instead of hardcoded lists.](/skills/neon-neon-ai-gateway/)[### Together AI ↗ Routes Together AI work to domain guides, with rules against retrying unrecoverable errors and leaving billable resources running.](/skills/togethercomputer-together-ai/)[### Amazon Bedrock ↗ Bedrock model invocation, Knowledge Bases, Guardrails and AgentCore guidance, with quota, prompt-caching and cost-tracking workflows.](/skills/aws-amazon-bedrock/) [AI Tools ↗](/tools/)[Skills ↗](/skills/)[Agents ↗](/agents/)[MCP Servers ↗](/mcp-servers/)[Workflows ↗](/workflows/) ## Continue your investigation - [Choose a toolkit](/tools/) - [Inspect execution hosts](/agents/) - [Check tool access](/mcp-servers/) - [Check project dependencies](/check-repo/)