Google ADK Engineering Profile
Review a Google Agent Development Kit design without confusing framework support with product authority. Read the ADK version boundaries reference when identifying an API, runtime skill, evaluation, session, or deployment contract.
Preconditions
Identify the repository, language, installed ADK version, intended users, runtime environment, model provider, registered tools, session service, memory service, artifact service, and decision under review. If any version-sensitive behavior is material, verify it against the documentation for the exact installed version.
Workflow
- Separate the developer assistant from the product agent. A skill that helps a person build an ADK application is not automatically a safe runtime bundle for a SaaS user.
- Inventory the agent hierarchy, models, instructions, callbacks, planners, tools, toolsets, sessions, state, memory, artifacts, retrieval, evaluation, and deployment boundaries actually present.
- Pin the language, ADK package version, model or provider configuration, and relevant experimental features. Record observed support separately for Python, TypeScript, and Go.
- Trace identity and authority from the product user through the host application to each agent and tool. Skill prose cannot create credentials, permissions, approvals, or a registered tool.
- If
SkillToolset is proposed, describe the runtime bundle, allowed content, integrity checks, release state, revocation, and progressive resource loading. Treat runtime Agent Skills as experimental until exact-version tests establish the required behavior.
- Compose with
agent-context-state-memory-design, agent-tool-action-boundary-review, agentic-system-security-review, retrieval-grounding-quality-review, agent-evaluation-design, and observability-design where their focused questions apply.
- Separate local architecture evidence, synthetic tests, provider evaluation, authentication, infrastructure creation, deployment, and production acceptance. This workflow does not authorize deployment or any external action.
- Report confirmed behavior, version-sensitive assumptions, missing evidence, and the smallest safe next test.
Boundaries
- Google Agents CLI is optional developer tooling, not an ADK runtime dependency or a source of product authority.
- Experimental tool confirmation is not durable approval for real external writes. A production design needs authenticated approvers, external decision state, idempotency, replay protection, and audit evidence.
- An ADK runtime bundle must not be downloaded from a mutable branch during a run.
- Framework availability does not establish security, evaluation quality, production readiness, or client parity.
- This skill does not authorize deployment, cloud access, authentication, billing, publication, or release.
Output
Return Decision, Exact versions, Developer and runtime planes, Agent composition, State and memory, Tool boundaries, Runtime skills, Evaluation, Security, Operational boundaries, Confirmed evidence, Experimental or version-sensitive behavior, Missing evidence, and Next safe test.
1---2name: google-adk-engineering-profile3description: Review or design a Google Agent Development Kit system using exact framework, language, model, session, tool, evaluation, and deployment evidence. Use for human-facing ADK architecture and version discovery; combine it with the focused agent security, evaluation, context, retrieval, and operational skills instead of treating framework defaults as proof of safety or readiness.4license: MIT5---67# Google ADK Engineering Profile89Review a Google Agent Development Kit design without confusing framework support with product authority. Read [the ADK version boundaries reference](references/adk-version-boundaries.md) when identifying an API, runtime skill, evaluation, session, or deployment contract.1011## Preconditions1213Identify the repository, language, installed ADK version, intended users, runtime environment, model provider, registered tools, session service, memory service, artifact service, and decision under review. If any version-sensitive behavior is material, verify it against the documentation for the exact installed version.1415## Workflow16171. Separate the developer assistant from the product agent. A skill that helps a person build an ADK application is not automatically a safe runtime bundle for a SaaS user.182. Inventory the agent hierarchy, models, instructions, callbacks, planners, tools, toolsets, sessions, state, memory, artifacts, retrieval, evaluation, and deployment boundaries actually present.193. Pin the language, ADK package version, model or provider configuration, and relevant experimental features. Record observed support separately for Python, TypeScript, and Go.204. Trace identity and authority from the product user through the host application to each agent and tool. Skill prose cannot create credentials, permissions, approvals, or a registered tool.215. If `SkillToolset` is proposed, describe the runtime bundle, allowed content, integrity checks, release state, revocation, and progressive resource loading. Treat runtime Agent Skills as experimental until exact-version tests establish the required behavior.226. Compose with `agent-context-state-memory-design`, `agent-tool-action-boundary-review`, `agentic-system-security-review`, `retrieval-grounding-quality-review`, `agent-evaluation-design`, and `observability-design` where their focused questions apply.237. Separate local architecture evidence, synthetic tests, provider evaluation, authentication, infrastructure creation, deployment, and production acceptance. This workflow does not authorize deployment or any external action.248. Report confirmed behavior, version-sensitive assumptions, missing evidence, and the smallest safe next test.2526## Boundaries2728- Google Agents CLI is optional developer tooling, not an ADK runtime dependency or a source of product authority.29- Experimental tool confirmation is not durable approval for real external writes. A production design needs authenticated approvers, external decision state, idempotency, replay protection, and audit evidence.30- An ADK runtime bundle must not be downloaded from a mutable branch during a run.31- Framework availability does not establish security, evaluation quality, production readiness, or client parity.32- This skill does not authorize deployment, cloud access, authentication, billing, publication, or release.3334## Output3536Return `Decision`, `Exact versions`, `Developer and runtime planes`, `Agent composition`, `State and memory`, `Tool boundaries`, `Runtime skills`, `Evaluation`, `Security`, `Operational boundaries`, `Confirmed evidence`, `Experimental or version-sensitive behavior`, `Missing evidence`, and `Next safe test`.
JovaniPink