waybarrios/opencode-power-pack

AI Slop Rubric defines observable signs of generic or poorly grounded interface design, with severity levels, evidence needs, and repair actions. Use it to review marketing sites, product pages, dashboards, portfolios, and e-commerce layouts.

AWS Context Discovery reads the local AWS profile, region, account, and caller identity before AWS or SageMaker work. Use it when cloud context is needed and you want to avoid guessing configuration.

Agentic Actions Auditor reviews GitHub Actions workflows that invoke AI agents for prompt injection, unsafe interpolation, sandbox gaps, and permissive actor rules. Use it when auditing agentic CI workflows and cross-file references in Actions configs.

Agents MD Revise captures recurring learnings from a session into AGENTS.md, CLAUDE.md, or a local override so future sessions have the same context. Use it when you need to remember project-specific commands, conventions, or gotchas.

Agents Md Improver audits AGENTS.md, CLAUDE.md, and related rules files for coverage, clarity, and freshness. Use it when project rules need review or targeted updates after code changes.

Code Architect analyzes an existing codebase’s patterns and conventions to produce an implementation blueprint for a non-trivial feature. Use it when planning architecture, file changes, component design, data flow, and build steps.

Code Explorer traces a feature through a codebase, from entry points and execution paths to supporting layers and dependencies. Use it to understand how existing code works before modifying or extending it.

Code Quality helps agents review code, run linting and type checks, and flag maintainability, complexity, and standards issues. Use it for quality gates and cleanup across Rust, TypeScript, Python, shell, and mixed repositories.

Code Review reviews a frozen set of code changes for bugs, logic errors, and convention violations. Use it when you need a PR or diff checked before merging.

Code Reviewer inspects small code changes for bugs, logic errors, security issues, and convention mismatches. Use it when reviewing a diff, a specific file, or a function and you want only high-confidence findings.

CodeQL Analysis runs CodeQL database creation, data extensions, and security queries, and can process SARIF output. Use it when CodeQL is explicitly requested or when you need deeper taint-flow security analysis.

Design Patterns helps agents choose implementation patterns and abstraction boundaries across Rust, TypeScript/React, and Django/Python. It covers when to use, when to avoid, and trade-offs for each option.

Differential Review performs security-focused reviews of PRs, commits, and diffs. It adapts depth to codebase size, uses git history and test coverage, and produces a markdown audit report.

False Positive Check systematically verifies a suspected security bug and returns a TRUE POSITIVE or FALSE POSITIVE verdict with evidence. Use it when you need to confirm whether a reported vulnerability is actually exploitable.

Feature Development guides a new feature through codebase understanding, clarifying questions, architecture design, implementation, and quality review. Use it when you need a methodical way to add functionality instead of jumping straight into code.

Frontend Design creates distinctive, production-grade web interfaces with accessible markup and strong visual craft. Use it when building or beautifying pages, components, dashboards, landing pages, or React and HTML/CSS UI.

HF Cloud Python Env Setup creates an isolated Python environment with a compatible Python version and boto3 for SageMaker or AWS automation. Use it before running AWS Python code; it is not for general Python setup.

HF Mem estimates the VRAM or memory needed to load Hugging Face Safetensors or GGUF model weights for inference. Use it to check whether a model fits on a GPU or instance, and optionally include KV cache estimates.

Hugging Face CLI manages Hub authentication, downloads, uploads, repositories, cache, jobs, buckets, webhooks, and endpoint administration from the command line. Use it for Hub CLI operations, not model selection or Space building.

Hugging Face Community Evals runs inspect-ai and lighteval evaluations for Hugging Face Hub models on local hardware. Use it to choose an inference backend, smoke-test tasks, and run GPU evals with vLLM, Transformers, or accelerate.

HuggingFace Best Model Finder finds recommended Hugging Face models for a task by checking official benchmark leaderboards and model size. It helps when you need to compare candidates against device constraints or choose a top model to try.

SageMaker Deployment Planner helps choose the right Amazon SageMaker hosting path for a model, including real-time or async inference. Use it to gather only the key deployment details before handing off to image, IAM, and endpoint setup skills.

SageMaker IAM Preflight checks for an existing SageMaker execution role before creating models, endpoints, or training jobs. It validates a provided role ARN or discovers a usable role when IAM access is failing.

SageMaker Production Defaults creates a real-time SageMaker endpoint with a model, endpoint config, autoscaling, and CloudWatch alarms. Use it after the image and IAM role are known to deploy production serving defaults and smoke-test the endpoint.

Serving Image Selection selects the correct SageMaker serving container URI for a model and region, then verifies it from the AWS catalog before deployment. Use it after choosing a deployment path and before endpoint code.