v1.4.0

@weisser-dev/awesome-opencode v1.4.0

Published to npm: @weisser-dev/awesome-opencode@1.4.0

npx @weisser-dev/awesome-opencode

---

20 Docker Dev Environments

Smart image selection: recommended images shown first based on your detected project languages.

Language	Image	Package Manager
Node.js 22	node:22	npm
Bun	oven/bun:latest	bun
Python 3.13	python:3.13-slim	pip
Python + uv	ghcr.io/astral-sh/uv:python3.13-bookworm-slim	uv
Java + Maven	maven:3-eclipse-temurin-21	maven
Java + Gradle	gradle:8-jdk21	gradle
Kotlin	openjdk:21-slim-bookworm + sdkman	gradle
Go 1.24	golang:1.24	go
Rust	rust:1-slim-bookworm	cargo
Ruby 3.3	ruby:3.3-slim	bundler
PHP 8.4	php:8.4-cli	composer
.NET 9 SDK	mcr.microsoft.com/dotnet/sdk:9.0	dotnet
Swift 6.1	swift:6.1	swift
Dart 3.7	dart:3.7	pub
Elixir 1.18	elixir:1.18-slim	mix
C/C++ (GCC 14)	gcc:14	make
Terraform	hashicorp/terraform:latest	terraform
Ansible	cytopia/ansible:latest	ansible
Ubuntu 24.04	ubuntu:24.04	apt — install anything via apt
Generic	node:22	npm (fallback)

Corporate Proxy & Artifact Registry (in sandbox mode)

Configure once, auto-injected into Docker container:

• HTTP/HTTPS corporate proxy
• Nexus npm registry + Maven mirror (auto-generates ~/.m2/settings.xml)
• JFrog Artifactory npm + Maven
• PyPI mirror for pip/uv

Unit Tests: 87 tests, 0 failures

npm test

Covers all data modules: agents, skills, models, fingerprinting, Docker images, MCP, languages.

Codebase Split into Modules

setup.js went from 2699 lines to 35 lines. All logic is now in src/data/ and src/lib/:

src/data/   — pure data (agents, skills, models, MCP, docker, languages)
src/lib/    — logic (detect, prompts, generate, docker sandbox)
src/tests/  — unit tests

See CHANGELOG.md for full details.

v1.3.0

@weisser-dev/awesome-opencode v1.3.0

Published to npm: @weisser-dev/awesome-opencode@1.3.0

npx @weisser-dev/awesome-opencode

New: CLI Flags for Enterprise Environments

Three new flags for corporate and custom setups:

--config <path> — Use an external opencode.json from anywhere:

awesome-opencode --config ~/shared/opencode.json
awesome-opencode --config /mnt/team-configs/opencode.json

• External configs are auto-mounted read-only into Docker sandbox
• Detection output shows the full path: Existing config: yes (/path/to/config.json)

--crt <path> — Custom CA certificate for TLS:

awesome-opencode --crt /etc/ssl/corporate-ca.crt

• Sets NODE_EXTRA_CA_CERTS for Node.js
• Certificate is mounted into Docker sandbox at /certs/ (read-only)
• Supports .crt and .pem files

--skipSSL — Disable TLS certificate verification:

awesome-opencode --skipSSL

• Sets NODE_TLS_REJECT_UNAUTHORIZED=0
• Useful behind corporate proxies with self-signed certificates

All flags can be combined and are automatically forwarded into Docker sandbox:

awesome-opencode --config /mnt/config/opencode.json --crt /mnt/certs/ca.pem --skipSSL

Fixes

• CLI banner renamed from "OpenCode Advanced Setup" to "Awesome OpenCode"
• README terminal output fully updated to match actual program prompts
• Docker mount changed from -v $(pwd):/workspace to -v $PWD:$PWD — preserves host paths inside the container
• All code blocks in README now use shell syntax highlighting

Model Fingerprints: 75 Patterns

Expanded from 29 to 75 model fingerprints covering all families from models.dev:

Anthropic, OpenAI (incl. GPT-4.1, GPT-5 mini, codex-mini), Google (incl. Gemini flash-lite), xAI/Grok (2/3/4), Cohere (Command-A/R), Perplexity (Sonar), Alibaba/Qwen, MiniMax, ZhipuAI/GLM, Nvidia/Nemotron, Moonshot/Kimi, Cerebras, StepFun, Inception/Mercury, and more.

See CHANGELOG.md for full details.

v1.2.0

@weisser-dev/awesome-opencode v1.2.0

Published to npm: @weisser-dev/awesome-opencode@1.2.0

npx @weisser-dev/awesome-opencode

See CHANGELOG.md for details.

Full Changelog: v1.1.1...v1.2.0

v1.1.1

@weisser-dev/awesome-opencode v1.1.1

Published to npm: @weisser-dev/awesome-opencode@1.1.1

npx @weisser-dev/awesome-opencode

Bug Fix

• Removed double start confirmation — When choosing "Start OpenCode" or "Start OpenCode (Sandboxed)" from the re-run menu, it no longer asks "Start OpenCode? Y/N" again. Direct launch.
• Sandboxed option shows Docker requirement — Menu entry displays "(Docker required)" and is greyed out when Docker is not installed, instead of silently failing.

What's in v1.1.x

New in v1.1.0:

• Sandboxed OpenCode via Docker — Run in isolated container with provider-aware env config (Anthropic, OpenAI, AWS Bedrock, Azure, Google, OpenRouter, Custom)
• CLI subcommands — awesome-opencode configure agents|skills|models|mcp for targeted changes
• Improved re-run menu — 8 options: Start / Sandboxed / Reconfigure / Configure agents/skills/models/mcp / Exit
• MCP search via mcp.so — Replaces broken registry API, supports multiple searches, filters mirrors
• Existing config detection — Shows custom providers, agents, and MCP servers found in your config
• GPT-5.3 fingerprints — Recognizes GPT-5.3, GPT-5.3 Codex, GPT-5.3 Chat

Full feature set (since v1.0.0):

• 108 agents across 10 categories
• 15 skills
• 18 curated MCP servers + mcp.so live search
• Smart model detection (26+ fingerprints, custom providers)
• Step limits per agent for cost control
• Language-aware agent recommendations
• AGENTS.md auto-generation

See CHANGELOG.md for full details.

v1.1.0

@weisser-dev/awesome-opencode v1.1.0

Published to npm: @weisser-dev/awesome-opencode@1.1.0

npx @weisser-dev/awesome-opencode

New Features

Sandboxed OpenCode via Docker:

• Run OpenCode in an isolated Docker container — only the current project is accessible
• Provider-aware env config for 7 providers: Anthropic, OpenAI, AWS Bedrock, Azure OpenAI, Google AI, OpenRouter, Custom
• Auto-detects existing env vars, shows Docker command for review before execution
• Recommended for enterprise setups

CLI Subcommands:

• awesome-opencode configure agents|skills|models|mcp — targeted changes without full reconfigure
• awesome-opencode --help — usage info

Improved Re-Run Menu:

• Start OpenCode / Start OpenCode (Sandboxed) / Reconfigure / Configure agents/skills/models/mcp / Exit

MCP Search via mcp.so:

• Replaces the unreliable registry.modelcontextprotocol.io API
• Supports multiple searches in a loop
• Filters out mirrors, asks for npm package confirmation

Existing Config Detection:

• Shows custom providers, existing agents, and MCP servers from your opencode.json

GPT-5.3 Model Fingerprints:

• Recognizes GPT-5.3, GPT-5.3 Codex, GPT-5.3 Chat from any provider

See CHANGELOG.md for full details.

v1.0.1

@weisser-dev/awesome-opencode v1.0.1

Published to npm: @weisser-dev/awesome-opencode@1.0.1

npx @weisser-dev/awesome-opencode

See CHANGELOG.md for details.

Full Changelog: v1.0.0...v1.0.1

v1.0.0 - Initial Release

awesome-opencode v1.0.0

The most comprehensive collection of agents, skills, and tooling for OpenCode — the open-source AI coding agent.

npx @weisser-dev/awesome-opencode

---

Highlights

108 Agents across 10 Categories

Category	Count	Examples
Core Development	8	api-designer, fullstack-developer, graphql-architect
Language Specialists	22	typescript-pro, python-pro, react-specialist, rust-engineer
Infrastructure	13	terraform-engineer, kubernetes-specialist, cloud-architect
Quality & Security	11	code-reviewer, penetration-tester, chaos-engineer
Data & AI	12	llm-architect, prompt-engineer, mlops-engineer
Developer Experience	11	mcp-developer, cli-developer, dx-optimizer
Specialized Domains	8	blockchain-developer, game-developer, iot-engineer
Business & Product	9	product-manager, scrum-master, ux-researcher
Meta & Orchestration	7	workflow-orchestrator, multi-agent-coordinator
Research & Analysis	7	trend-analyst, competitive-analyst

Inspired by VoltAgent/awesome-claude-code-subagents, fully adapted for OpenCode with permissions, skills, and markdown agent format.

15 Skills

git-release, pr-review, migration, test-patterns, deploy, dependency-audit, incident-postmortem, docker-optimize, adr-write, api-contract, changelog-generate, ci-pipeline, env-setup, error-triage, performance-profile

Smart Model Detection

Recognizes models from any provider — including custom Bedrock, Azure, or self-hosted endpoints:

my-provider/eu.anthropic.claude-opus-4-6-v1  → claude-opus-4  | Frontier | $$$$ | Coding: 95/100
my-provider/eu.anthropic.claude-haiku-4-5    → claude-haiku-4 | Fast     | $$   | Coding: 75/100

• 26 model fingerprints across Anthropic, OpenAI, Google, DeepSeek, Meta, Mistral
• Auto-optimize option uses YOUR existing models for best frontier/fast mix
• Agent-tier mapping: which agents need frontier vs. strong vs. fast models
• Based on pricepertoken.com benchmarks

Step Limits & Cost Control

Every agent has a recommended step limit:

• Code-writing agents: unlimited
• Review/analysis agents: 10-15 steps
• Fast/read-only agents: 5-10 steps
• Applied automatically via CLI prompt

18 Curated MCP Servers + Live Registry Search

Curated servers filtered by your languages, plus live search via the official MCP Registry API (registry.modelcontextprotocol.io).

Intelligent Setup Flow

• Fresh vs existing project detection
• Multi-select languages with file-extension scanning (auto-detected shown first)
• Language-aware agent recommendations (e.g. Terraform → terraform-engineer, platform-engineer, sre-engineer)
• AGENTS.md auto-generation with language-specific conventions
• Re-run detection via .opencode/advanced.json — "Start / Reconfigure / Exit"
• Auto-launch OpenCode after setup

Documentation

7 comprehensive docs covering agents, skills, MCP servers, models, rules, tools, and permissions.

---

Installation

# One command — analyzes your project, sets everything up, launches OpenCode
npx @weisser-dev/awesome-opencode

# Or install globally
npm install -g @weisser-dev/awesome-opencode
awesome-opencode

---

Links

• Repo: https://github.com/weisser-dev/awesome-opencode
• npm: https://www.npmjs.com/package/@weisser-dev/awesome-opencode
• OpenCode: https://opencode.ai
• Inspiration: VoltAgent/awesome-claude-code-subagents