opencode-power-pack addresses a thorny problem in AI coding agent ecosystems: how to reuse complex multi-agent workflows designed for one platform (Anthropic’s Claude Code) within another (OpenCode). Claude Code plugins put their logic in commands/ and agents/ directories, which OpenCode cannot natively interpret. This pack reverse-engineers those workflows and translates them into OpenCode-compatible SKILL.md files, preserving the original methodologies without rewriting the logic from scratch.
What opencode-power-pack does and how it works
At its core, opencode-power-pack ports eleven official Claude Code skills into OpenCode format. These skills cover a range of AI coding assistant workflows including code review with confidence-filtered cross-checks, security audits categorized by OWASP buckets (requiring concrete proofs of concept for attacks), a structured seven-phase feature development pipeline, frontend design generation suitable for production, MCP server authoring, and project memory management with AGENTS.md tooling.
The challenge is architectural: Claude Code plugins organize their logic into commands/ and agents/ directories, which OpenCode cannot process. Instead of running these plugins as-is, opencode-power-pack rewrites these multi-agent workflows as native SKILL.md files, the format OpenCode understands for defining skills. This translation acts as a Rosetta Stone between two AI coding ecosystems.
Technically, the pack is implemented in JavaScript, relying on OpenCode’s runtime environment that bundles Bun (a JavaScript runtime) to manage dependencies and execution. The project pairs well with the obra/superpowers plugin, which offers meta-workflow skills that extend OpenCode’s capabilities with domain-specific muscle.
Installation requires just one config line to include the plugin, plus symlinking slash commands from a local clone or GitHub source. This minimal setup reflects a tradeoff: while the translation enables reuse, you still need to maintain a local copy of the repo for the slash command files, since they are not embedded in the plugin itself.
Why the translation layer is the technical strength — and its tradeoffs
What sets opencode-power-pack apart is its focus on bridging two incompatible plugin architectures. Claude Code’s multi-agent workflows are complex, relying on a directory structure and execution model that OpenCode does not support. By reverse-engineering these workflows and rewriting them as SKILL.md files, the pack preserves the original agents’ logic and coordination patterns.
This approach is more maintainable than reimplementing the skills from scratch in OpenCode, and it respects the original design and methodology. The code is organized clearly around the translation process, with a focus on fidelity to the source workflows rather than reinventing them.
The tradeoff is that the pack depends on symlinks to local slash command files, which means that while the skill logic is native to OpenCode, some components remain external. This can complicate updates and offline use, although the repo supports two installation paths: one from GitHub and one from a local clone. The local clone installation avoids network access at runtime and lets you tinker or contribute more easily.
The code quality reflects a pragmatic approach: it avoids overengineering, favors clarity, and leverages OpenCode’s built-in JavaScript environment. The pack does not attempt deep integration or runtime reflection of Claude Code internals; rather, it focuses on translating static workflows into a compatible format.
Quick start: installing opencode-power-pack
Installation requires OpenCode on your system, plus git for fetching the plugin. Bun is bundled with OpenCode, so dependency installation happens automatically.
Pick an installation method:
A — Install from GitHub
Add the plugin to your OpenCode config (~/.config/opencode/opencode.json):
{
"$schema": "https://opencode.ai/config.json",
"plugin": [
"opencode-power-pack@git+https://github.com/waybarrios/opencode-power-pack.git"
]
}
If you use other plugins like superpowers, include them all in the array:
{
"plugin": [
"superpowers@git+https://github.com/obra/superpowers.git",
"opencode-power-pack@git+https://github.com/waybarrios/opencode-power-pack.git"
]
}
Pin a specific release version if desired:
"opencode-power-pack@git+https://github.com/waybarrios/opencode-power-pack.git#v0.1.0"
Next, clone the repo locally (required for slash commands):
git clone https://github.com/waybarrios/opencode-power-pack.git ~/code/opencode-power-pack
B — Install from a local git clone
For offline use or development, clone the repo locally and add the plugin pointing to the local path. This method avoids GitHub access at runtime.
Both methods require symlinking the slash command files and restarting OpenCode to activate the skills.
verdict
opencode-power-pack is a niche but solid solution for anyone who wants to leverage Anthropic’s Claude Code skills within OpenCode. The translation layer is its key technical strength, enabling reuse of complex multi-agent workflows without rewriting them from the ground up.
The tradeoff is that its setup involves managing local symlinks and keeping a clone of the repo handy, which may add friction for some users. The pack does not encapsulate everything purely within OpenCode’s plugin model, but this is understandable given the fundamental architectural mismatch.
For developers working with AI coding agents, prompt engineering, or multi-agent orchestration, this repo is worth exploring. It provides a concrete example of cross-platform skill translation and highlights challenges in evolving AI assistant ecosystems. If you are invested in OpenCode and want to tap into Claude Code’s workflows, this pack is the bridge you need.
Related Articles
- Inside Claude Code: A detailed reconstruction of Anthropic’s AI safety and architecture — A deep dive into Claude Code’s 512K lines of TypeScript reveals a layered YOLO safety classifier, multi-agent IPC, and t
- Inside Claude Code From Scratch: A practical reconstruction of Anthropic’s coding agent — Claude Code From Scratch distills Anthropic’s 500K+ line coding agent into ~8,000 lines of Python and TypeScript, reveal
- claude-code-harness: a Shell-based plugin harness for Claude Code AI agents — claude-code-harness is a Shell plugin harness for Claude Code that integrates AI agent features without Node.js, relying
- Softaworks Agent Toolkit: A modular plugin marketplace for AI coding agents — Softaworks Agent Toolkit offers 40+ modular AI skills and plugins for coding agents like Claude Code, enabling composabl
- Claude Code Projects Index: the agent workspace pattern for domain-specific Claude Code agents — Claude Code Projects Index is an Astro-based curated index of 75+ Claude Code agent workspaces. It uses the agent worksp
→ GitHub Repo: waybarrios/opencode-power-pack ⭐ 351 · JavaScript