opencode-config
Portable AI coding assistant config with memory & subagent pipeline
❓ Why
Moving your AI coding assistant config to a new server meant setting everything up from scratch every time. Plus no versioning — break a skill and you can't rollback to what worked.
✅ What
A ready-to-run Docker config. Clone, fill .env, start — it works. Versioning via Git, memory syncs through a private repository.
features
- Portable: Docker config — runs on any server with one command
- Versioning: Git under the hood — rollback to any config version
- Global skills: All skills mounted via volume — work across all projects
- Portable memory: Markdown + git, syncs through private repo — assistant remembers everything
- Dual search: Keyword (ripgrep) always works, semantic — optional via OpenAI API
- Pipeline: ISSUE → IMPLEMENT → DOCS → CI → REVIEW → MERGE → MEMORY
- Subagent orchestration: Orchestrator only plans, subagents execute
- Linear execution: One pipeline at a time — no branch conflicts
- Auto-issue: Bug found mid-task — agent creates GitHub issue and continues
- Skills & commands: Built-in skills and slash commands, main one — `/run-pipeline`
- Subagents: Built-in role-based agents — review, docs, memory sync