gca-americas

A comprehensive virtual TA skill for the "Building Trustworthy Charity Agents with Google ADK and AP2" codelab. This skill provides deep, grounded guidance for a 120-minute workshop covering multi-agent architectures, the Agent Payments Protocol (AP2) credential chain (IntentMandate → CartMandate → PaymentMandate), Google ADK orchestration patterns (Agent, SequentialAgent, FunctionTool, callbacks), and production deployment to Vertex AI Agent Engine and Cloud Run with Cloud Trace observability.

A comprehensive virtual TA skill for the "ADK Agentic Pattern with Memory & MCP" codelab. Covers 7 sessions building progressively complex ADK agent architectures — from single agents through Sequential, Parallel, Loop, Custom (BaseAgent), Routing, Agent-as-Tool, Long-Term Memory (DatabaseSessionService), and MCP Toolbox integration.

A comprehensive virtual TA skill for "Agentverse - The Shadowblade's Codex - Vibecoding with Gemini CLI". Covers mastering Gemini CLI (commands, MCP servers, extensions, skills, hooks), vibe coding a Shadowblade combat agent with ADK, agent evaluation (adk eval + pytest), and building a CI pipeline with Cloud Build for containerized deployment to Cloud Run.

A comprehensive virtual TA skill for "Build a Secure Agent with Model Armor and Identity". Covers building a production-grade secure AI agent using Google ADK with enterprise security patterns — Model Armor for input/output filtering, Agent Identity for least-privilege BigQuery access control via conditional IAM, OneMCP for BigQuery data access, deployment to Agent Engine, and red team attack validation.

A skill that provides ADK Crash Course - From Beginner To Expert workshop information based on reference data.

A skill that provides credit claim/redeem information and FAQ for the workshop.

A skill that provides Level 0 workshop information based on reference data.

A skill that provides Level 1 workshop information based on reference data.

Transform a raw product idea, brief, or rough technical sketch into a sharp software design doc written by a senior staff engineer who has shipped real systems. Use this skill whenever the user mentions a "software design doc", "engineering design doc", "eng spec", "technical design doc", "TDD", "architecture doc", "system design", "implementation plan", or says anything like "how do we build this", "tech-spec this", "what's the architecture", "scope the engineering", "the eng side of this", "design the backend". Also triggers when the user has bullet-point thoughts about data models, APIs, infra, or trade-offs and wants them expanded. Output is a single self-contained Markdown design doc covering architecture (with a mermaid diagram), data model, API surface, key trade-offs with rejected alternatives, risks, a testing strategy (consumed by the `test-driven-dev` skill), rollout plan, and an honest scope of what's NOT being built. Push toward boring tech, smallest viable architecture, and explicit trade-offs —

Transform a rough product idea, hackathon pitch, or napkin sketch into a sharp product design doc (PRD) written by a senior PM who actually ships. Use this skill whenever the user mentions a "product idea", "PRD", "product spec", "product design doc", "product brief", "feature pitch", a "hackathon project", or says anything like "I want to build X" / "turn this into a real product doc" / "let's PM this" / "what's the spec for this" / "make this shippable" — even if they don't say "PRD" explicitly. Also triggers when the user has a bullet-point pitch (contract, scope, magical moment, signature detail) and wants it expanded. The output is a single self-contained Markdown PRD with sharp scope discipline, opinionated trade-offs, and a clear handoff to UX and Eng. Push toward shipping the smallest thing that proves the magic — not the org chart's wishlist. This skill operates independently — it reads only the raw brief, not other design docs.

Transform a raw product idea, brief, or rough pitch into a sharp UX design doc written by a senior product designer who actually ships. Use this skill whenever the user mentions a "UX spec", "UX design doc", "design doc", "interaction design", "wireframes", "user flow", "screen-by-screen design", or says anything like "design this", "what does it look like", "spec the UI", "the UX side of this", "make this designable". Also triggers when the user has bullet-point thoughts about screens, flows, components, or interactions and wants them expanded. Output is a single self-contained Markdown UX design doc covering the user journey, screen-by-screen specs, interaction details, edge states, and the one signature interaction that defines the product. Push toward fewer screens, fewer states, one defining interaction — not a Figma library of every possible component. This skill operates independently — it reads only the raw brief, not other design docs.

Guide for adding coins (treats) to the score collection and creating announcements in Firestore. Use when the user wants to reward users with treats or broadcast a message.

Streamlined instructions for diagnosing issues within the DinoQuest application environment using Google Cloud MCP servers. Use when the user asks "Find out what's wrong with dinoquest" or needs a quick server health check.

Diagnose and remediate Cloud Run service errors — OOM, crash loops, bad deployments, and misconfiguration.

Agentic CI pipeline for DinoQuest. Reads the GitHub PR context to understand what changed, scopes tests intelligently (frontend-only, backend-only, or full), builds the Docker image via Cloud Build, pushes to Artifact Registry, interprets failures with LLM reasoning, and posts a commit status + PR comment back to GitHub.

Autonomous canary deployment for DinoQuest. Risk-scores the change, sets canary %, monitors Cloud Run metrics, and promotes or rolls back. Learns from past deployments via Firestore memory — recognized patterns skip the slow ramp (Behavior A).

Sets up a Cloud Logging Log Router sink to stream live Cloud Run server logs into a BigQuery dataset, then queries BigQuery via MCP to generate a quick traffic and error report, and a game analytics report with feature insights. Use when asked to route Cloud Run logs to BigQuery, set up log export, analyze server logs, generate a log report from BigQuery, or surface feature ideas from game event data.

Sets up a Cloud Logging Log Router sink to stream live Cloud Run server logs into a BigQuery dataset, then queries BigQuery via MCP to generate a quick traffic and error report, and a game analytics report with feature insights. Use when asked to route Cloud Run logs to BigQuery, set up log export, analyze server logs, generate a log report from BigQuery, or surface feature ideas from game event data.