OpenCode Configuration

Forked From: https://github.com/flpbalada/my-opencode-config

My personal OpenCode configuration optimized for full-stack web development with autonomous agent workflows.

Feel free to:

• Browse and adapt skills for your own config
• Use agents as templates for your workflows
• Share tips and tricks you've found useful

Structure

.
├── AGENTS.md          # Personal preferences & coding guidelines
├── opencode.json      # Main config (plugins, MCP servers, providers)
├── agents/            # Custom agents for specialized tasks
├── skills/            # executable knowledge notes
├── commands/          # Custom slash commands
└── scripts/           # Helper scripts 

Custom Agents

Agent	Purpose
architect	High-level planning and architecture (Tier 1)
coder-monitor	Autonomous build validation (Tier 3)
code-reviewer	Deep quality and security audit (Tier 4)
web-researcher	Internet research and documentation
requirements-analyzer	Feature requirements analysis
deep-thinker	Structured thinking for complex problems
effort-estimator	Estimate development effort
refactoring	Code restructuring and improvement
commit-drafter	Generate conventional commit messages
code-simplifier	Simplify code for clarity
skill-creator	Create new skills with proper structure
chat	Conversational interactions

Skills Library

Skills are executable knowledge notes — a collection of patterns, best practices, and frameworks converted into actionable guidance for the AI agent.

Category	Skills
Full-Stack Web	React + Vite + Redux, Node + PostgreSQL + InfluxDB, Python FastAPI/Django
Desktop Apps	Tauri 2.0 + Rust, .NET 9 Minimal APIs, Electron Secure IPC
Cloud & DevOps	AWS/Azure/GCP Deployment, Git Worktree Management
GIS Development	PostGIS + Leaflet
TypeScript	best practices, advanced types, satisfies operator, interface vs type
React	useState, useCallback, key prop, "use client" boundaries
CSS	container queries, Tailwind v4 best practices
Architecture	naming conventions, project structure, wrong abstraction patterns
Product Frameworks	Jobs-to-be-Done, Business Model Canvas, Hooked Model, Fogg Behavior Model, PEST analysis, product decisions
UX Psychology	cognitive load, cognitive biases, cognitive fluency, Hick's law, progressive disclosure, trust signals, halo effect
Behavioral Design	loss aversion, status quo bias, social proof, curiosity gap, self-initiated triggers, visual cues & CTAs
Decision Making	hypothesis trees, five whys, graph thinking, game theory (tit-for-tat)
Agile	Kanban, theme-epic-story hierarchy, user stories
Product Management	what not to do as PM

MCP Servers

• Context7 — Real-time library/API documentation

Plugins

• @mohak34/opencode-notifier — Desktop notifications
• @franlol/opencode-md-table-formatter — Markdown table formatting
• opencode-wakatime — Time tracking
• @tarquinen/opencode-dcp — Context management

Commands

Command	Purpose
/implement	Execute implementation following architect's plan
/git-commit	Create git commit with conventional messages
/learn	Extract patterns from session to create new skills
/summarize-branch	Generate PR description from commits

Resources

• OpenCode Documentation
• Skills Guide
• Agents Guide

License

MIT