{"id":43903936,"url":"https://github.com/apenlor/opencode-expert-mode","last_synced_at":"2026-02-06T19:24:58.436Z","repository":{"id":333757839,"uuid":"1135294572","full_name":"apenlor/opencode-expert-mode","owner":"apenlor","description":"A comprehensive configuration suite for agentic development, offering structured workflows and best practices.","archived":false,"fork":false,"pushed_at":"2026-01-29T07:46:31.000Z","size":302,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-01-29T22:49:36.661Z","etag":null,"topics":["agentic-ai","ai-agents","developer-tools","expert-mode","llm-tools","opencode","productivity","software-engineering","workflow"],"latest_commit_sha":null,"homepage":"","language":"TypeScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/apenlor.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-01-15T22:43:50.000Z","updated_at":"2026-01-26T16:42:04.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/apenlor/opencode-expert-mode","commit_stats":null,"previous_names":["apenlor/opencode-expert-mode"],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/apenlor/opencode-expert-mode","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apenlor%2Fopencode-expert-mode","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apenlor%2Fopencode-expert-mode/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apenlor%2Fopencode-expert-mode/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apenlor%2Fopencode-expert-mode/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/apenlor","download_url":"https://codeload.github.com/apenlor/opencode-expert-mode/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/apenlor%2Fopencode-expert-mode/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":29173671,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-02-06T16:33:35.550Z","status":"ssl_error","status_checked_at":"2026-02-06T16:33:30.716Z","response_time":59,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic-ai","ai-agents","developer-tools","expert-mode","llm-tools","opencode","productivity","software-engineering","workflow"],"created_at":"2026-02-06T19:24:57.998Z","updated_at":"2026-02-06T19:24:58.416Z","avatar_url":"https://github.com/apenlor.png","language":"TypeScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"[![Codacy Badge](https://app.codacy.com/project/badge/Grade/228a9a530e5f4f7fafd9cad5e66d2413)](https://app.codacy.com/gh/apenlor/opencode-expert-mode/dashboard?utm_source=gh\u0026utm_medium=referral\u0026utm_content=\u0026utm_campaign=Badge_grade)\n[![Latest Tag](https://img.shields.io/github/v/tag/apenlor/opencode-expert-mode)](https://github.com/apenlor/opencode-expert-mode/releases/latest)\n[![License](https://img.shields.io/badge/License-MIT-blue.svg)](https://opensource.org/licenses/MIT)\n\n# OpenCode Expert Mode\n\nThis repository provides an advanced agent configuration for OpenCode, designed to create a powerful and reliable software engineering assistant. It is a native OpenCode adaptation of the original [Superpowers for Claude](https://github.com/obra/superpowers) project by @obra.\n\n## Table of Contents\n- [Global Installation](#global-installation)\n- [Verify Installation](#verify-installation)\n- [Development](#development)\n- [Basic Workflow](#basic-workflow)\n- [Core Philosophy](#core-philosophy)\n- [Components](#components)\n  - [Agents](#agents)\n  - [Skills](#skills)\n  - [Commands](#commands)\n  - [Plugins](#plugins)\n- [Directory Structure](#directory-structure)\n\n## Global Installation\n\nThis configuration is intended to be installed globally by cloning it directly into your OpenCode configuration directory. This makes the tools and skills available across all your projects.\n\n### Prerequisites\n- [OpenCode CLI](https://opencode.ai) installed and available in your PATH.\n\n### 1. Back Up Your Existing Configuration\n**IMPORTANT**: This will prevent you from overwriting any custom setups you may have.\n```bash\nmv ~/.config/opencode ~/.config/opencode.bak\n```\n\n### 2. Clone the Repository\nClone this repository directly into the `~/.config/opencode` directory.\n```bash\ngit clone git@github.com:apenlor/opencode-expert-mode.git ~/.config/opencode\n```\n\n### 3. Set Up Your Local Configuration\nThis repository provides different example configuration files for various providers and setups. Choose the one that best fits your needs and copy it to `opencode.json`.\n\n```bash\ncd ~/.config/opencode\n# Choose one of the following:\ncp opencode.geminicli.example.json opencode.json   # For Google Gemini (Recommended)\n# cp opencode.github.example.json opencode.json    # For GitHub Models\n# cp opencode.antigravity.example.json opencode.json # For Antigravity\n# cp opencode.custom.example.json opencode.json    # For custom provider setups\n\n# Also copy the agents configuration template\ncp AGENTS.example.md AGENTS.md\n```\nYou can now safely customize `opencode.json` and `AGENTS.md` without creating conflicts with future updates from this repository.\n\n## Verify Installation\n\nTo ensure the configuration is correctly loaded:\n\n1.  **Start a new OpenCode session:**\n    ```bash\n    opencode\n    ```\n2.  **Ask the agent about its mode:**\n    \u003e What mode are you in?\n\n    It should confirm that it is in \"Expert Mode.\" This verifies the plugin is loading the core skill.\n\n3.  **Test a Command:** Ask the agent to plan a simple task using a command.\n    ```\n    /write-plan \"create a hello world script in python\"\n    ```\n4.  **Confirm Behavior:** The agent should respond by confirming it is using the `writing-plans` skill to create the plan. This verifies that the commands and skills are working together correctly.\n\n## Development\n\n### Debugging the Plugin\nThe core plugin (`plugins/expert-mode-plugin.ts`) includes a detailed logging mechanism for development. By default, these logs are sent to the main OpenCode log files.\n\nTo get a dedicated, clean log file in the project root, you can enable a special debug mode by setting an environment variable before launching OpenCode:\n\n```bash\nexport EXPERT_MODE_DEBUG=1\nopencode\n```\n\nWhen enabled, a `plugin-debug.log` file will be created in the root folder where you're executing opencode. This file contains detailed logs from the Expert Mode plugin only, making it much easier to trace its behavior during development.\n\n## Basic Workflow\n\nThis configuration enables a structured, expert-guided development lifecycle using the provided commands.\n\n1.  **Design (`/brainstorm`):** Start by exploring an idea to solidify requirements.\n    ```\n    /brainstorm \"a web server that returns the current time\"\n    ```\n2.  **Plan (`/write-plan`):** Generate a detailed, step-by-step implementation plan in the chat.\n    ```\n    /write-plan \"a simple python flask server with one endpoint /time\"\n    ```\n3.  **Execute (`/execute-plan`):** Instruct the agent to begin implementing the plan from the chat context.\n    ```\n    /execute-plan\n    ```\n4.  **Review (`@code-reviewer`):** After work is complete, call the specialized code reviewer for feedback.\n    ```\n    @code-reviewer Please review the flask server implementation.\n    ```\n\n**Note:** While these commands are convenient shortcuts, the skills are the true core of this configuration. They are designed to be used by *any* agent, enhancing its ability to reason and execute tasks effectively, regardless of how it's invoked.\n\n## Core Philosophy\n\nThe central idea of Expert Mode is a **\"Skill-as-Core\"** architecture.\n\n-   **Skills (`skills/`)**: The heart of the project. They contain expert workflows that enhance any agent's ability to perform complex tasks.\n-   **Commands (`commands/`)**: A user-facing \"control panel\" that provides convenient shortcuts to directly invoke specific skills.\n-   **The Agent**: The agent is empowered by this ecosystem. Whether responding to a general prompt or a specific command, it can use its `skill` tool to access these expert workflows at any time.\n\n## Components\n\nThis configuration is composed of several key components that work together.\n\n### Agents\n- **`code-reviewer`**: A subagent designed for in-depth code reviews. Invoke with `@code-reviewer`.\n- **`spec-reviewer`**: Reviews an implementation against a specification.\n- **`implementer`**: Implements a single, well-defined task from a plan.\n\n### Skills\nA collection of expert workflows in the `skills/` directory. Key skills include:\n- **`brainstorming`**: A structured process for exploring ideas and refining them into concrete designs (presented in-chat).\n- **`dispatching-parallel-agents`**: For tackling multiple independent tasks at once.\n- **`executing-plans`**: A systematic way to execute implementation plans with review checkpoints.\n- **`finishing-a-development-branch`**: For guiding the completion and integration of development work.\n- **`receiving-code-review`**: For processing and implementing feedback from a code review.\n- **`requesting-code-review`**: For verifying work meets requirements before merging.\n- **`subagent-driven-development`**: For executing implementation plans with independent tasks.\n- **`systematic-debugging`**: A disciplined process for identifying and resolving bugs.\n- **`test-driven-development`**: A guide for writing tests before implementation code.\n- **`using-git-worktrees`**: For creating isolated git worktrees for feature work.\n- **`using-expert-mode`**: Establishes how to find and use skills (this is the core skill loaded on session start).\n- **`verification-before-completion`**: For running verification checks before claiming work is complete.\n- **`writing-plans`**: A TDD-centric approach to creating detailed, bite-sized implementation plans (presented in-chat).\n- **`writing-skills`**: For creating, editing, and verifying new skills.\n\n### Commands\nUser-facing shortcuts in the `commands/` directory that invoke skills.\n- **`/brainstorm`**: Kicks off the `brainstorming` skill.\n- **`/write-plan`**: Starts the `writing-plans` skill.\n- **`/execute-plan`**: Begins the `executing-plans` skill.\n\n### Plugins\nThe plugin in `plugins/expert-mode-plugin.ts` is the entry point that bootstraps the agent into Expert Mode. It uses modern OpenCode hooks to ensure the agent's core identity is always present.\n\n- **`experimental.chat.system.transform`**: On every chat turn, this hook injects the `using-expert-mode` skill directly into the system prompt. This makes the agent's expert identity impossible to forget.\n- **`experimental.session.compacting`**: When a long conversation is summarized, this hook adds a flag to the summary, ensuring the \"Expert Mode\" state persists across context compactions.\n\n## Directory Structure\n\nThis repository's root is designed to be your OpenCode configuration directory.\n```\n.\n├── AGENTS.example.md   # A template for your local agent rules.\n├── agents/             # Definitions for specialized subagents (e.g., code-reviewer).\n├── commands/           # User-facing slash commands that invoke skills.\n├── opencode.example.json # An example configuration for user-specific settings (e.g., models).\n├── plugins/             # OpenCode plugins that extend core behavior (e.g., session hooks).\n└── skills/              # The core skills that define expert workflows.\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapenlor%2Fopencode-expert-mode","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fapenlor%2Fopencode-expert-mode","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fapenlor%2Fopencode-expert-mode/lists"}