
Adk Builder
- 3 repo stars
- Updated January 18, 2026
- MattMagg/agentic-plugins
adk-builder is an agent skill that runs spec-driven autonomous workflows for Google ADK agent development.
About
adk-developer is a Prism agent skill for and developers who ship on Google’s Agent Development Kit (ADK). It packages autonomous, spec-driven workflows so your coding agent does not freestyle directory layout, agent roles, or integration steps every time you start a new ADK feature. Use it when you have—or are writing—a concrete spec for an ADK agent, tool surface, or multi-agent flow and want repeatable procedural guidance that keeps implementation aligned with that spec. The skill sits primarily on the Build journey shelf under agent-tooling, but it naturally touches Validate when you are still scoping what the agent must do before you commit to code. It is a skill package for agentic development discipline, not a hosted MCP server or a generic Python snippet pack. Pair it with your normal repo, cloud credentials, and ADK docs; review the Security Audits panel on this catalog page before enabling shell or network-heavy steps in untrusted environments.
- Spec-driven autonomous workflows tailored to Google ADK agent projects
- Guides end-to-end ADK development from structured requirements into implementation
- Fits solo builders using Claude Code, Cursor, or Codex as the coding agent
- Centers multi-agent and tool-using patterns common in ADK stacks
- Reduces ad-hoc prompting when standing up or extending ADK-based agents
Adk Builder by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
/plugin marketplace add MattMagg/agentic-plugins/plugin install adk-builder@agentic-pluginsAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| repo stars | ★ 3 |
|---|---|
| Last updated | January 18, 2026 |
| Repository | MattMagg/agentic-plugins ↗ |
What it does
Install adk-developer when you want a spec-driven, autonomous workflow for scaffolding and extending Google ADK agent projects without improvising each step in chat.
Who is it for?
Best when you're already committed to Google ADK and want repeatable, spec-first agent implementation in an existing repo with a capable coding agent.
Skip if: Skip if you're building non-agent SaaS with no ADK, or anyone and already has a frozen spec and only needs a single-line code edit without an ADK workflow.
When should I use this skill?
When you are developing or extending Google ADK agents and want autonomous, spec-driven workflows rather than unstructured chat implementation.
What you get
You get a spec-aligned ADK development path your agent can follow autonomously, producing structured agent project changes instead of one-off snippets.
- ADK-aligned agent and tool implementation changes in the repo
- Spec-consistent project structure for ongoing ADK work
- Documented next implementation steps tied to the spec
Files
Autonomous Google ADK development with spec-driven workflows
adk-builder
Source: ./plugins/adk-builder
{ "name": "adk-builder", "source": "./plugins/adk-builder", "category": "development", "description": "Autonomous Google ADK development with spec-driven workflows" }
Related skills
How it compares
Use instead of generic “build me an agent” prompts when you specifically want Google ADK structure and spec-driven steps, not a language-agnostic planner skill alone.
FAQ
Who is adk-builder for?
adk-developer is for developers shipping AI agents on Google ADK who want their Claude Code, Cursor, or Codex session to follow a consistent spec-driven development workflow rather than improvising each file and integration.
When should I use adk-builder?
Use adk-developer during Validate when you are locking agent scope and tool boundaries into a spec, during Build when you are implementing ADK agents and orchestration, and during Ship when you are hardening or extending an ADK stack you already started with the same spec-driven pr
Is adk-builder safe to install?
Treat adk-developer like any agent skill that may suggest shell, network, or cloud API work: read what it will run in your repo, scope credentials minimally, and review the Security Audits panel on this Prism page rather than relying on invented audit scores.