Skip to content

v0.3.0

Choose a tag to compare

@github-actions github-actions released this 28 Sep 19:24
· 156 commits to main since this release

gemini-api has been rewritten in Go. This release replaces the TypeScript CLI shipped in v0.1.0–v0.2.1 and is not backward compatible: commands, flags and output formats have changed.

Install

macOS / Linux:

curl -fsSL https://raw.githubusercontent.com/google-gemini/gemini-api-cli/main/scripts/install.sh | bash

Windows (PowerShell):

iwr -useb https://raw.githubusercontent.com/google-gemini/gemini-api-cli/main/scripts/install.ps1 | iex

Go:

go install github.com/google-gemini/gemini-api-cli/cmd/gemini-api@latest

Or download an archive below (Linux, macOS, Windows · x86_64, arm64) and verify it against checksums.txt.

What's new

Dedicated commands per task. Capabilities that were flags on run are now their own commands:

  • generate: text and multimodal generation
  • image: image generation and editing
  • tts: text to speech, including multi-speaker
  • analyze: questions about images, audio, video, documents or YouTube URLs
  • New: music, video, transcribe, tokens

Interaction lifecycle.

  • agent run --background returns an interaction ID immediately.
  • agent status, agent cancel and agent delete-interaction manage it afterwards.

New API coverage.

  • files: Files API upload, list, get, delete, and registering Cloud Storage objects
  • models: curated catalog plus live list/get
  • environments, triggers, webhooks, credentials
  • docs: Gemini API docume
    Setup and auth.
  • configure, auth and whoami store credentials in ~/.config/gemini-api/config.yaml or the OS keyring.
  • GEMINI_API_KEY still works.
  • OAuth via --access-token / GEMINI_ACCESS_TOKEN, and quota projects via --user-project.

Scripting and agents.

  • -o pretty|json|yaml|table|toon, --jq, --dry-run
  • --body / --schema for full request control, and --usage for a machine-readable CLI schema
  • --agent-mode for structured errors
  • Documented exit codes: 0 ok · 1 runtime · 2 usage · 3 auth

Interactive. explore browses and runs commands, and completion supports bash, zsh, fish and PowerShell.

Changelog

  • 23c8d31 fix(cli): set project_name: gemini-api in .goreleaser.yaml
  • 552828c Merge pull request #38 from google-gemini/import-test-new
  • 77848e2 Update the test
  • 2eae3b1 First import of the test/* and evals/*
  • 895be4d fix(cli): remove unconfigured voices service from CLI surface
  • e855807 feat(cli): bind -i, --id on triggers run and triggers list-executions
  • 148e3b7 chore: Add licensing and disclaimer section to README files.
  • c31a119 chore(cli): add model suggestions for image, video, and music commands in CLI overlay
  • 7a2e8ce chore(cli): regenerate CLI with Speakeasy v2.941.1
  • 20aaa63 chore(cli): override environments and environments files list help text in CLI overlay
  • f5168fd Merge pull request #23 from google-gemini/copybara/987626002
  • 896a7b5 fix(cli): align stale path keys in CLI overlays with GAOS OpenAPI spec
  • d9b392e chore(cli): improve webhook examples in CLI overlay
  • af53a80 chore: Add licensing and disclaimer section to README files.
  • d3c649a No public description
  • cb63496 Merge pull request #14 from google-gemini/feat/add-yaml-secrets-var
  • ed136c6 feat(config): resolve env placeholders in agent yaml
  • b2c15ec test(config): add env placeholder fixture