OpenCode Power Pack plugin
55 skills for Claude Code, Codex, OpenCode and Pi: code review, security audit, feature development, frontend design and Hugging Face
- 546 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The code-review skill posts comments on a GitHub pull request when asked
- The SageMaker and Hugging Face Jobs skills create cloud resources and can spend money
- The skills run Semgrep, CodeQL and commands in the repository working copy
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/opencode-power-packDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The code-review skill posts comments on a GitHub pull request when asked
- The SageMaker and Hugging Face Jobs skills create cloud resources and can spend money
- The skills run Semgrep, CodeQL and commands in the repository working copy
This is third-party code. Review the repository files before installing.
What it does
OpenCode Power Pack bundles 55 portable skills that work in Claude Code, Codex, OpenCode and Pi. Most of them cover review and security: a multi-agent code-review for pull requests with confidence filtering, security-review, threat modeling, Semgrep and CodeQL runs, SARIF parsing, insecure-default detection and dependency risk checks. For development there is a seven-phase feature-dev with the code-explorer, code-architect and code-reviewer helpers, plus code quality and design pattern skills. The second large block is Hugging Face: the hf CLI, model discovery by leaderboards, fine-tuning with TRL, Sentence Transformers and vision models, Gradio Spaces and deployment on AWS SageMaker. The set also holds frontend-design with an accessibility rubric, ai-slop for judging generic design, humanizer for English and Spanish copy, mcp-builder, skill-creator and audits of project rule files. You can install it whole as a plugin or selectively through npm with profiles such as recommended, review, security and huggingface. Since version 0.5.0 there is an optional sandboxed command runner: it uses Seatbelt on macOS and Bubblewrap on Linux.
Who it is for. For developers and ML engineers who work across several agents and want one skill set for review, security and model training.
Good fit when
- You need a pull request reviewed through several lenses with confidence filtering and reproduction scenarios
- You need a repository threat model and Semgrep or CodeQL runs with SARIF parsing
- You want a feature taken from code exploration to implementation and review in seven phases
- You want to fine-tune a model with TRL or Sentence Transformers and publish a demo as a Gradio Space
- You want only the review skills, without the ML catalog, through the review profile
Not a fit when
- You want a development process with brainstorming, TDD and plans: the authors point to superpowers, which this pack complements
- You want a small install of a few skills: the full plugin loads all 55
- You work on native Windows without WSL2: the sandbox is not ready there and stops execution
- You need a single-file review: the pack has a separate code-reviewer for that
Example request
/opencode-power-pack:code-review review pull request 128 and show only confirmed issuesLimitations
The sandbox is optional and needs Node.js 20.11.0 or newer. Linux needs bubblewrap, socat and rg, and it does not work on WSL1 or native Windows. The stated isolation level is shell-contained: the command and its descendants are protected, not the whole agent. The SageMaker and Hugging Face skills need cloud accounts and permissions. With the full plugin installed, a selective copy of the same skills does not shrink the loaded catalog.
How to disable. In Claude Code disable or remove the plugin with /plugin. In Codex run codex plugin remove opencode-power-pack@opencode-power-pack, in Pi pi remove git:github.com/waybarrios/opencode-power-pack, and delete selectively copied skills from ~/.agents/skills. Remove the global npm package with npm uninstall --global @waybarrios/opencode-power-pack.
FAQ
How do I install OpenCode Power Pack in Claude Code?
Run /plugin marketplace add waybarrios/opencode-power-pack and /plugin install opencode-power-pack@opencode-power-pack. Skills get the plugin prefix, for example /opencode-power-pack:code-review.
How do I install only the review skills?
With npx @waybarrios/opencode-power-pack install --profile review. There are also profiles for feature-dev, frontend, security, huggingface, authoring and project-memory. The --dry-run flag previews a selection without writing to disk.
How is code-review different from code-reviewer?
Code-review checks a pull request, branch or commit range through several lenses and can post validated findings to GitHub when asked. Code-reviewer takes one file, function or small local change and works lighter.
What does the sandbox exec runner do?
It runs a command inside Seatbelt isolation on macOS or Bubblewrap on Linux with one of four profiles: observe, develop, network-read, publish. If isolation is unavailable, the command does not run.
Does the pack work in Pi?
Yes. The command pi install git:github.com/waybarrios/opencode-power-pack installs the package and pi list shows it. The -l flag saves the install in project settings.
Discussion
No questions or reviews yet.
Related
PluginsA modular plugin set for Claude Code: skills, agents, commands and rules for Java, TypeScript, Python, PHP and AWS
Eleven skills for Claude Code and Codex: test audit, git worktree cleanup, plan pages, screen recording transcripts and 1Password sign-in
OpenCode plugin that turns a session into a team of agents led by an architect and runs the work through plan phases
A Claude Code and Codex plugin with skills for code research, tests, refactoring and independent review of finished changes
A code-review CLI and plugin for agents: reads a git diff and writes line-level comments with context
Claude Code Skills and Agents
Claude Code Skills & Agents
A Claude Code plugin bundling skills and sub-agents: deep research, image generation, GitHub work, code review and model switching
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Repository stars | Main difference |
|---|---|---|---|
| OpenCode Power Pack (this entry) | no | 546 | |
| Superpowers | no | 307,845 | provides a development process of brainstorming, plans, TDD and subagents; Power Pack complements it with review, security and ML |
| Trail of Bits skills | no | 7,867 | Trail of Bits plugin marketplace for security audits |
README in short
The page opens with the version 0.5.0 sandbox, then gives quick installs for Claude Code, Codex and Pi, a selective npm install with profiles and a table of sandbox profiles. All skills follow by category, with a table for choosing between code-review and code-reviewer. It closes with requirements, per-agent install checks, updating and credits to the upstream authors. MIT and Apache-2.0 licenses.