GoogleCloudPlatform

Compares a modernized Next.js application against its legacy Express counterpart using runtime side-by-side verification. Use when ensuring functional and business logic parity between two systems.

Manages the end-to-end modernization of legacy Express monoliths into Next.js architectures. Orchestrates subagents for auditing, scaffolding, and verification. Use when starting or managing a greenfield rewrite project.

Systematically probe a modernized Next.js application for logic flaws, security vulnerabilities, or missing features. Use this to find bugs or cases where the migration failed to match legacy behavior.

Analyzes Express route definitions and controller logic to document API endpoints, payloads, and response structures. Use when reverse-engineering an existing Express application's API surface.

Analyzes authentication flows, authorization rules, middleware logic, and side-effects. Use when extracting business rules, Passport configurations, or mailer logic from an Express application.

Analyzes legacy ORM models (Mongoose, Sequelize) to extract schemas, validations, and relationships. Use when reverse-engineering a legacy data layer for a modern rewrite.

Analyzes legacy UI templates (Pug, EJS, HTML) to extract a comprehensive inventory of components, layouts, and conditional logic. Use when reverse-engineering a legacy frontend for a modern rewrite.

Reads API_Contracts.md and generates failing Vitest integration tests for the expected Next.js Route Handlers. Use when generating API tests for TDD.

Expert instructions for building high-quality GKE troubleshooting skills. Codifies Step 0 context rules, zero-hallucination signatures, and explicit LQL/PromQL query requirements.

Diagnose and prevent `vbar_control_agent` segfaults and OOMs caused by race conditions during TPU device resets and frequent metrics collection (e.g. every 3s). Use when TPU slice initialization fails or `vbar_control_agent` crashes on TPU v6e nodes.

Assists in preparing applications and clusters on GKE for production.

Workflows for containerizing and deploying applications to GKE for the first time.

Workflows for auditing and hardening the security of GKE workloads.

Answer natural language questions about GKE-related costs by leveraging BigQuery export and cost allocation data.

Guides the user through creating GKE clusters using pre-defined templates (Standard, Autopilot, GPU/AI).

Guidance on managing the lifecycle and upgrades of Google Kubernetes Engine (GKE) clusters.

Use when the marathon plan needs non-spatial event parameters defined: theme, hydration-point logistics, crowd zones, start times, or expected participant counts. Triggered when the request describes the feel or operational logistics of the race rather than its physical route.

Use when the marathon plan needs a 26.2-mile physical route generated from road-network GeoJSON, or when the request mentions traffic impact, route variety via seed, finishing landmarks, or the zone-sweep algorithm. Triggered by route-planning or traffic-analysis language.

Use when the planning committee asks about marathon finances or requests a budget change. Triggered by mentions of budget, cost, spending, sponsorship, or any phrasing that implies modifying a financial allocation. Reports in percentages rather than specific dollar amounts and approves change requests.

Use when planning a marathon route that must reference real-world geography: looking up landmarks, businesses, or parks in a city, computing distances between points, or checking historical and forecasted weather for the race date and location. Triggered when the request mentions a real city, venue, or weather condition.

Use when the planning committee asks to change, increase, decrease, or modify a budget allocation in a security-restricted context. Triggered by any budget-modification phrasing. Refuses the change and directs the requester to an authorized administrator.

Use when a runner agent receives a distance update and must deplete water proportionally, or when the runner enters a hydration station and must decide whether to stop and rehydrate. Triggered by distance events and station events.

Use when a runner agent needs to manage athletic performance during a race tick: accelerating, braking, or reading current vitals (speed, energy, hydration, exhaustion). Triggered every tick the runner is active.

Use when the simulator's loop agent advances the race by one tick: broadcasting state to runners, draining the telemetry buffer, computing traffic conditions, and checking for race completion. Triggered every iteration of the LoopAgent.

Plan a remediation strategy after a test failure.

Executes code generation from Magic Modules into downstream Terraform providers and optionally compiles the binary.

Triggers the QA Test Runner subagent to reproduce failures and interpret logs.

Checks sync status and invokes the subagent to align repositories if needed.

Gather context on a change or bug and plan the implementation.

Checks if the downstream provider repository is aligned with the correct Magic Modules base commit.

Configures Terraform CLI with developer overrides to test locally built provider binaries without full Go acceptance tests, enabling rapid plan/apply cycles.

Parses massive Terraform debug logs into an easy-to-read outline of API delays, failures, and request payloads.

Guidelines and rules for provisioning Google Cloud Spanner instances, including selecting the correct edition (Enterprise vs Standard) and compute capacity based on schema features (e.g., graphs and vector search) and environment purpose.

Google Ads copywriting expertise. Covers Responsive Search Ad (RSA) best practices, headline formulas, description frameworks, and character limits. Load this skill when generating text ads, headlines, or ad copy.

Brand strategy and campaign positioning expertise. Covers audience segmentation, competitive differentiation, value proposition frameworks, and campaign brief creation. Load this skill when setting up campaigns, selecting audience segments, or defining creative direction.

Ad platform specifications and best practices. Covers Google Ads, YouTube, Instagram, TikTok, and Meta ad formats, dimensions, character limits, and optimization tips. Load this skill when creating platform-specific content or recommending campaign settings.

Market trend analysis and product-trend mapping expertise. Covers how to interpret trend data, align products to cultural moments, and create trend-informed campaign angles. Load this skill during the Creative Research Phase when running trend_spotter and product_trend_mapper.

Cinematic video ad storytelling expertise. Covers 3-act narrative structure, emotional arcs, camera motion, scene transitions, and voiceover pacing for 24-second commercial video ads. Load this skill when generating video ads or storylines.

Visual art direction expertise for marketing imagery. Covers photography styles, composition rules, color theory, lighting techniques, and product staging. Load this skill when generating image ads, asset sheets, moodboards, or keyframe images.

Google Ads copywriting expertise. Covers Responsive Search Ad (RSA) best practices, headline formulas, description frameworks, and character limits. Load this skill when generating text ads, headlines, or ad copy.

Moves a mockgcp service from using locally generated grpc-gateway proto bindings to using reflection with the official Google Cloud Go client library (httptogrpc). Use this when tasked with stopping grpc-gateway generation for a GCP service in mockgcp.

Creates or updates the _identity.go and _reference.go files for a Config Connector resource, ensuring they follow the canonical gcpurls.Template pattern. Use this when you need to make sure the identity and reference is up to date for a KCC resource or when implementing IdentityV2 and refs.Ref for a resource.

Implement the controller, mappers, and fuzzer for a direct KCC resource, ensuring package isolation and CI compliance. Use this when implementing the main reconciliation logic for a "direct" resource.

Guides the implementation of KRM types and CRD scaffolding for new "direct" resources.

Provides instructions and a tool for analyzing Custom Resource Definition (CRD) changes, including checking for KRM API structural equivalence and backward compatibility.

Checks apis/ subdirectories for missing generate.sh and helps create PRs to add them, following the pattern in

Provides a structured logic for capturing and routing agentic learnings to prevent knowledge pollution.

Implement the core reconciliation logic (Adapter) and E2E fixtures for a direct KCC resource. Use this after Step 1 is complete.

Installs Google's GenMedia MCP servers (Lyria, NanoBanana, Veo, Chirp, AVTool) via curl from pre-compiled release binaries and registers them in mcp_config.json. Use when the required media synthesis tools are missing or inactive.

Expert in generating full multi-scene multimedia storybooks (image, video, voice, and music) with dynamic duration probing, conversational tempo guardrails, a dedicated self-correcting Editor's QC Room, and pipeline flowcharts embedded in interactive reports.

Builds the mcp-genmedia Go MCP servers (nanobanana, veo, lyria, gemini-multimodal, chirp3-hd, avtool) from source and wires them into settings.json. Use this skill whenever the MCP tools are missing or broken — typically at the start of a new session, after a container restart, or when /tmp has been wiped. The prebuilt binaries in /workspace/.local/bin/ have no exec bit and live on a noexec mount; this skill compiles fresh executables into /tmp/bin/ where execution is allowed.

Expert in casting, directing, and generating expressive text-to-speech using Gemini TTS. Use this when the user needs virtual voice actor personas, expressive speech generation, or multiple variations of a voiceover (like "take 3 on the bounce").

Expert media production assistant. Use when requested to help with storyboarding, podcast creation, audio assembly, or complex multi-step media workflows using the GenMedia MCP servers (Veo, Lyria, Gemini TTS, NanoBanana).

Expert in video composition, editing, and format conversion. Use when the user wants to generate high-quality video, overlay images on video, concatenate clips, create GIFs, or sync audio to video using mcp-avtool-go and mcp-veo-go.

Expert in audio synthesis, music generation, and mixing. Use when creating podcasts, background scores, or multi-track audio layering using mcp-chirp3-go, mcp-lyria-go, mcp-gemini-go, mcp-nanobanana-go, and mcp-avtool-go.

Expert in AI image generation and editing. Use when the user needs high-quality textures, character-consistent visuals, or image-to-image editing using mcp-nanobanana-go.

End-to-end GECX/CXAS/CES conversational agent lifecycle -- build agents from requirements (PRD-to-agent), create and run evals (goldens, simulations, tool tests, callback tests), debug failures, and iterate to production quality. Use this skill whenever the user mentions GECX, CXAS, CES, SCRAPI, conversational agents, voice agents, audio agents, agent evals, pushing/pulling/linting agents, or agent instructions/callbacks/tools on the Google Customer Engagement Suite platform.

Migrate Dialogflow CX (DFCX) agents to CXAS (Customer Experience Agent Studio) agents. Use this skill when the user mentions DFCX migration, migrating agents, converting DFCX to CXAS, porting agents, agent migration, or post-migration optimization/consolidation. Four independently runnable scripts: migrate.py (1:1), stage_1.py (variable dedup + consolidation), stage_2.py (instruction state machines + tool mocks + lint + report), stage_3.py (rewires consolidated topology from source dep graph; only needed when stage_1 ran consolidation). State persists between scripts via <target>_ir.json so each can run / re-run / resume independently.

Retrieves non-contained CCAI Insights conversations (losses), uses agent intelligence to cluster them into common failure patterns, and generates a markdown report. Use when you need to analyze failure patterns and build targeted regression/evaluation reports.

Automates the ingestion of customer requirement documents such as diagrams, BRDs, code etc., synthesizes high-fidelity natural transcripts, and compiles them into highly interactive, responsive Critical User Journey (CUJ) reports.

A robust methodology for LLM-based requirements gathering and high-fidelity artifact generation. Employs 'Divide, Conquer, and Verify' tactics using specialized subagents, iterative exhaustion loops, and batched execution to ensure zero data loss.

A two-phase protocol for extracting structure and generating transcripts from customer artifacts.

Enforces task coverage and prevents drift via a deterministic checklist tool with output verification.

Extracts conversational transcripts from .drawio XML files.

Specialized in working with Google Maps Grounding lite and MCP tools. Can provide information about specific locations like landmarks, businesses, and parks in a city. Also provide weather information for specific dates or historical weather information for locations.

Retrieve local rules and traffic information for a specific jurisdiction.

TODO - One line description of what this skill does.

Evaluates the plan against financial, community, and logistical goals.

Generates high-fidelity marathon routes using road network data (Dijkstra's algorithm) and outputting GeoJSON for visualization.

Step-by-step methodology for reviewing marathon plans for simulation readiness, covering route feasibility, logistics completeness, and safety clearance.

Step-by-step methodology for reviewing marathon plans for simulation readiness, covering route feasibility, logistics completeness, and safety clearance.

Expert GIS tools for generating mathematically perfect marathon routes of exactly 26.2 miles using road network data and zone-sweep decomposition.

Guides the agent to bootstrap an initial context set (templates & facets) by deducing key information from the database schema and generating a ContextSet file.

Generate and expand datasets of Natural Language Questions (NLQ) and SQL pairs for evaluation.

Guides the agent to execute an evaluation of a generated ContextSet against a golden dataset utilizing the Evalbench framework.

Guides the agent to perform hill-climbing iterations to improve a ContextSet based on evaluation results.

Orchestrates the initialization workflow for auto context generation, and provides helper workflow for setting up dataset connection by creating or updating tools.yaml configurations.

Guidelines and best practices for generating context items (Templates, Facets, Value Searches). Use this skill whenever the user asks to create, author, or generate context for database enrichment, or asks for examples and instructions on how to write templates, facets, or value searches. It helps bridge the gap between LLMs and structured databases.

Build a competitor pipeline view for a target / mechanism / indication — who is in the clinic, what phase, what differentiation. Use for BD / portfolio / commercial-strategy questions like "who else is developing X" or "what's the pipeline for indication Y".

Surface emerging safety signals for a drug or class from PubMed (case reports / letters / RCT AE tables), plus relevant trial AE disclosures. Use for pharmacovigilance triage, medical-affairs safety updates, or competitive risk assessment. NOT a substitute for formal PV systems.

Produce a literature-grounded mechanism-of-action explanation for a drug or drug class — receptor / pathway / downstream effects / clinical relevance — with a Nano Banana Pro diagram and citations. Use for medical-affairs MSL prep, internal training, or when the user asks "how does drug X work" or "what's the MoA of class Y".

Translate a clinical or research question into a PICO/PECO-structured PubMed + Europe PMC search strategy with MeSH terms, field tags, and search hedges. Use whenever the user asks a comparative-effectiveness, etiology, prognosis, diagnosis, or HEOR question, OR when they explicitly ask for a "search strategy".

Run a PRISMA 2020-aligned systematic-review workflow — identification → screening → eligibility → included — with a transparent record of exclusions at each step and a rendered PRISMA flow diagram. Use when the user asks for a systematic review, evidence map, scoping review, or any task that requires a defensible screening audit trail.

Build a target-validation dossier for a gene / protein / pathway — biology, disease association, druggability, existing programs, key publications, trial pipeline, safety signals. Use for early-stage discovery target review, portfolio decisions, or when the user asks "what do we know about target X".

Guide for cutting a new release of the Cloud Foundation Fabric (CFF) repository. Use this skill when asked to create, prepare, or draft a new release.

Guides the user step-by-step through the prerequisites for the FAST 0-org-setup stage, supporting both Standard GCP and Google Cloud Dedicated (GCD) environments. Use when a user asks to prepare or run prerequisites for 0-org-setup or bootstrap the FAST landing zone.

A simple skill to test if the agent correctly executes tools.

A wizard to help users configure FAST (Fabric Architecture Setup Tool) step-by-step. Use this skill when asked to configure FAST, run the FAST wizard, or setup FAST.

Generates idiomatic Cloud Foundation Fabric (CFF) Terraform code using CFF modules. Use when users ask to create GCP resources, use Fabric modules, or generate Terraform code for Google Cloud.