YT · YOUTUBE// 37d agoPRODUCT UPDATE

Gemini CLI adds SDK, browser agent

Gemini CLI’s v0.30 and v0.31 releases turn Google’s open-source terminal agent into a more extensible platform, adding an SDK with custom skills support, project-level policy controls, experimental direct web fetch, an experimental browser agent, and Gemini 3.1 Pro Preview support. This is a meaningful upgrade for developers who want a terminal-first agent they can customize and govern more tightly.

// ANALYSIS

Google is pushing Gemini CLI past “AI in terminal” novelty and toward a real agent platform with extensibility, browser automation, and tighter policy guardrails.

–The new SDK and custom skills support make Gemini CLI more than a chat wrapper, giving teams a path to build reusable internal tooling on top of it
–Experimental browser agent support expands the product from code and shell tasks into web workflows, which is where many practical automation use cases live
–Project-level policy features and richer policy matching matter for enterprise adoption because terminal agents need clear controls before teams trust them with real repos
–Gemini 3.1 Pro Preview support keeps the CLI aligned with Google’s latest model stack, which helps it compete with fast-moving terminal agents from Anthropic and OpenAI

// TAGS

gemini-clicliagentsdkopen-sourceautomation

DISCOVERED

37d ago

2026-03-06

PUBLISHED

37d ago

2026-03-06

RELEVANCE

9/ 10

AUTHOR

AICodeKing