No description
  • TypeScript 100%
Find a file
2026-03-14 20:25:45 -05:00
agents Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00
commands Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00
plugins Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00
prompts Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00
skills Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00
README.md Initial commit of distilled oh-my-opencode knowledge 2026-03-14 20:25:45 -05:00

Oh-My-OpenCode Distilled

This repository contains distilled knowledge extracted directly from the Oh-My-OpenCode source code. The core concepts—system prompts, prompt snippets, plugins, and skill configurations—have been programmatically parsed and reformatted natively into ready-to-use OpenCode assets.

Directory Structure

  • agents/: Pure Markdown representations of Agent system prompts. Instead of digging through source code, you can read the raw system instructions that drive agents like Atlas, Oracle, Prometheus Planner, and the Discipline Agents orchestrator.
  • skills/: Contains the raw SKILL.md files used by the system (e.g., /init-deep, ultrawork, LSP + AST-Grep). These define the tools, instructions, and workflows for specific developer tasks and can be dropped directly into your workspace.
  • prompts/: Contains extracted prompt snippets and context-injection loops (like the Ralph Loop, Todo Enforcer, Comment Checker, and IntentGate). These can be used to augment your own custom system prompts.
  • plugins/: Contains TypeScript plugin definitions for advanced architectural features (like Hash-Anchored Editing, Tmux Integration, Background Agents, and Built-in MCP loaders). These demonstrate how to hook into the OpenCode lifecycle.

Features Converted to Assets

The 15 features from the Oh-My-OpenCode system have been directly converted into their respective native formats:

  1. Discipline Agentsagents/discipline-agents.md
  2. Prometheus Planneragents/prometheus-planner.md
  3. ultrawork / ulwskills/ultrawork.md
  4. LSP + AST-Grepskills/lsp-ast-grep.md
  5. /init-deepskills/init-deep.md
  6. Ralph Loopprompts/ralph-loop.md
  7. Todo Enforcerprompts/todo-enforcer.md
  8. Comment Checkerprompts/comment-checker.md
  9. IntentGateprompts/intentgate.md
  10. Hash-Anchored Edit Toolplugins/hash-anchored-edit.ts
  11. Background Agentsplugins/background-agents.ts
  12. Built-in MCPsplugins/built-in-mcps.ts
  13. Tmux Integrationplugins/tmux-integration.ts
  14. Claude Code Compatibleplugins/claude-code-compatible.ts
  15. Skill-Embedded MCPsplugins/skill-embedded-mcps.ts

How to Use This Distillation

  1. Viewing Agent Logic: Open any file in the agents/ directory to read the exact system prompt texts that shape the agent's persona and constraints.
  2. Reusing Prompts: Look in the prompts/ directory to see how OpenCode orchestrates loop-continuation and context compaction.
  3. Installing Skills: Copy any SKILL.md from the skills/ directory directly into your own project's .opencode/skills/ to instantly inherit those toolchains.
  4. Developing Plugins: Explore the plugins/ directory to see examples of the OpenCode Plugin API in action, utilizing hooks like tool.execute.before and onLoad.

Commands

The distilled/commands/ directory contains slash-command implementations (/<command>) for all features, demonstrating how to invoke them interactively within an OpenCode CLI session.