Use gplay with AI Coding Agents

gplay is the AI-agent-native Google Play Console CLI. Every design decision — minified JSON output, explicit long flags, --help on every command, --dry-run on every write, zero interactive prompts — exists so an AI coding agent can drive the entire Google Play workflow unattended.

Supported AI agents and frameworks

gplay works with every major AI coding agent and autonomous framework in 2026:

Agent / Framework	Type	Skills-compatible
Claude Code	AI coding CLI (Anthropic)	✅ Native
Cursor	AI-first IDE	✅ Native
Codex	AI coding CLI (OpenAI)	✅ Native
Gemini CLI	AI coding CLI (Google)	✅ Native
Windsurf	AI-first IDE	✅ Native
Aider	AI pair-programming CLI	✅ via prompt
Cline	VS Code agent	✅ Native
Continue	VS Code / JetBrains agent	✅ Native
GitHub Copilot CLI	AI shell assistant	✅ via prompt
Amazon Q Developer	AWS coding agent	✅ via prompt
OpenClaw	Autonomous agent runtime	✅ Native
Hermes Agent	Learning-loop agent framework (Nous Research)	✅ Native

Any agent that can read --help output and parse JSON can drive gplay, so the list above is not exhaustive.

Install the Agent Skills

Ready-made skills teach your agent the real Google Play workflows — release flow, staged rollouts, IAP setup, subscription base plans and offers, purchase verification, vitals monitoring, review management, screenshot automation, testers orchestration, metadata sync, migration from Fastlane, and pre-submission checks:

npx skills add tamtom/gplay-cli-skills

Source: tamtom/gplay-cli-skills. Sixteen skills total, all validated against the live CLI surface by a CI guardrail so they can never drift from the real commands.

Why agents work well with gplay

• Self-documenting — gplay --help, gplay tracks --help, gplay tracks update --help. Agents discover the interface instead of hallucinating it.
• Token-efficient JSON — minified by default; agents parse it directly. --output table exists for humans.
• No interactive prompts — destructive operations take a --confirm flag instead of blocking on stdin.
• --dry-run everywhere — agents can propose and preview writes safely before executing.
• Explicit flags — always --package, never -p; unambiguous for generation.

Example prompts

Things you can ask your agent once gplay is installed and authenticated:

Release this AAB to the internal track, then promote to beta once it’s live.

Create monthly and yearly subscriptions with a 7-day free trial, and convert prices for all regions.

Summarize this week’s crash clusters and file an issue for the top three.

Reply to all 1-star reviews from the last week that mention login problems.

Download last month’s earnings report.

CI / headless use

export GPLAY_SERVICE_ACCOUNT=/secrets/play-sa.json
export GPLAY_PACKAGE=com.example.app
export GPLAY_NO_UPDATE=1

gplay release --track internal --bundle app.aab

All configuration works via environment variables — no config file or browser required.