The skill requires enabling the workflow skill from the same repository first. Do not write code if the project is not scaffolded; if code already exists, run agents-cli scaffold enhance. Reference materials are divided into Python and Go: core APIs, graph-based workflows, real-time voice/video agents, as well as recipes for sandboxed code execution, cross-session memory, approval before dangerous operations, per-user credentials, scheduled and event triggers, etc.; for deeper questions, consult the documentation index at adk.dev.
Why it matters Organizes ADK agent types, tool definitions, callbacks, state, and graph-based workflows into quick-reference tables and a recipe index, so agents check recipes before writing code.
skills.sh installs 589036 (+865 / 24h)
Security Audit: Not audited
- Not yet scanned
Rationale
- skills.sh installs in 24 hours: 865, cumulative total: 589036 (represents attention only)
- SKILL.md is signed by Google, Apache-2.0; scaffolding and deployment are handled by other skills in the same repository, with clear division of labor
Usage
- First run uv tool install google-agents-cli to install the command-line tool, then run npx skills add https://github.com/google/agents-cli --skill google-agents-cli-adk-code, and tell the agent "Use ADK to write an agent that requires human approval before executing dangerous operations"
Risks
- Not audited: not scanned on desktop, website does not display installation commands
- Depends on separately installed agents-cli and the workflow skill from the same repository; the command-line tool itself is outside the scope of this review; deployment often goes through Google Cloud, which incurs cloud costs