An installable CoffeeMix extension pack for OpenCode + OMO, bundling specialist agents, local skills, validation tooling, and setup guidance.
coffeemix_all requires two programs installed on your system:
| Program | Description | Install |
|---|---|---|
| OpenCode | Terminal-based AI coding agent | npm install -g opencode-ai |
| oh-my-openagent | OpenCode plugin — multi-agent orchestration layer | npm install -g oh-my-opencode |
After installing both, run bunx oh-my-opencode install to configure your model providers.
Choose the path that matches your current setup:
- Fresh install: use the default installer path below. It is additive and skip-only for existing files.
- CoffeeMix 0.1 to 0.2 update: do not rely on the default installer. Use the explicit CoffeeMix-owned update mode described in
docs/UPDATE.md.
The default installer path keeps OMO / oh-my-openagent as the control plane and does not replace existing OMO routing, model, skill, or provider configuration.
Always preview first:
# Clone or download this package, then:
cd opencode_coffeemix_all_0_2
.\install-global.ps1 -WhatIfFor safe mock validation, use -ConfigDir:
$mock = "C:\temp\coffeemix-0-2-opencode-config"
.\install-global.ps1 -WhatIf -ConfigDir $mockOnly run the installer without -WhatIf after reviewing the output and backing up your global OpenCode config.
See docs/installation-guide.md for full installation options and docs/UPDATE.md for the 0.1 to 0.2 update path.
CoffeeMix 0.2 installer work is designed around explicit update flags:
$backup = "$env:USERPROFILE\.config\opencode-backups\coffeemix-0-1-before-0-2"
.\install-global.ps1 -WhatIf -UpdateCoffeeMix -BackupDir $backup
.\install-global.ps1 -UpdateCoffeeMix -BackupDir $backupIf your local install-global.ps1 does not list -UpdateCoffeeMix and -BackupDir, stop and use the default path only for fresh additive installs. A normal install skips existing CoffeeMix files, so it may leave old 0.1 agents, skills, or plugin files in place.
For update rehearsals, use -WhatIf -ConfigDir <mock OpenCode config> with a mock directory that contains OMO config, such as oh-my-opencode.json or oh-my-openagent.json. The update guide includes backup, rollback, and post-update validation commands.
coffeemix_all provides:
- 14
gb-*specialists — narrow-scope agents for review, debugging, planning, diagnostics, status, handoff, config, plugins, worktrees, and more - 8 local skills — reusable workflow gates for planning, debugging, verification, adoption intake, and completion checks
- Validation harness — scenario-based testing for both task behavior and specialist routing
- Global installer — PowerShell merge installer with OMO/OpenAgent config detection,
-WhatIf,-ConfigDir, plugin skip behavior, and JSON write protection during preview mode. Fresh installs are additive and skip existing files. 0.1 users need explicit CoffeeMix-owned update mode. - UI badge plugin — a CoffeeMix marker for OpenCode TUI environments
- Focused release docs — installation guide and release notes only
- Stable CoffeeMix-first
gb-*overlay while keeping OMO / oh-my-openagent as the control plane - Safer global installer guidance with mock-config validation support via
-ConfigDirand a documented explicit 0.1 to 0.2 update path - Documented local skills with
SKILL.mdfiles - CoffeeMix UI badge plugin included by default
- Release docs pruned to deployment essentials
- Explicit non-goals: no OMO replacement, no copied router/model/loader, no default GBrain/browser/checkpoint/design runtime, and no
gb-*deletion
| Path | Purpose |
|---|---|
.opencode/agents/ |
14 gb-* specialist prompts |
.opencode/skills/ |
8 local workflow skills with SKILL.md docs |
.opencode/plugins/ |
UI badge plugin |
assets/ |
Logo and visual assets |
docs/ |
Installation guide, update guide, and release notes |
scenarios/ |
Task-based behavior scenarios |
routing-scenarios/ |
Identity-prompt routing validation scenarios |
tools/ |
Validation harness and runners |
| Skill | Purpose |
|---|---|
ask-user-question |
Gate dangerous or irreversible operations behind explicit confirmation |
compact-context |
Summarize long-running sessions while preserving critical information |
enter-plan-mode |
Switch to planning mode before non-trivial implementation |
integration-intake |
Screen external workflow/tool adoption before integration |
systematic-debugging |
Root-cause-first investigation discipline |
test-driven-development |
Define red states or acceptance targets before changes |
tool-search |
Discover and select tools based on current task context |
verification-before-completion |
Require fresh evidence before declaring work complete |
Run the validation harness to verify specialist routing and scenario behavior:
# Smoke checks (file existence, inventory, subset scenarios)
python tools/workspace_smoke_runner.py
# Full e2e validation (all scenarios, strict checks)
python tools/workspace_e2e_runner.py
# Routing validation (identity-prompt scenarios across specialists)
python tools/routing_validation_runner.pyAlso verify JSON files parse:
Get-Content package.json -Raw | ConvertFrom-Json | Out-Null
Get-Content opencode.json -Raw | ConvertFrom-Json | Out-Null
Get-Content oh-my-opencode.json -Raw | ConvertFrom-Json | Out-Null
Get-Content tui.json -Raw | ConvertFrom-Json | Out-NullReports are written to reports/ and should not be committed as release artifacts.
This project builds on the work of others in the AI agent ecosystem. We are grateful to:
- OpenCode — The terminal-based AI coding agent runtime that powers this extension pack.
- oh-my-openagent — The multi-agent orchestration plugin that provides the OMO control plane.
- obra/superpowers — Workflow discipline and skill-authoring influence.
- garrytan/gstack — Specialist layering and optional adapter inspiration.
| Prefix / Source | Origin |
|---|---|
gb-* specialists |
Original to coffeemix_all ("gb" = 가베, old Korean word for coffee) |
| Local discipline skills | Adapted from obra/superpowers workflow gate concepts |
integration-intake |
Gate for CoffeeMix integration screening |
Baseline agents (sisyphus, oracle, librarian, etc.) |
Provided by oh-my-openagent |
All specialist prompts, local skills, validation scripts, and routing scenarios in this repository are independently authored workspace-local artifacts.
This project is licensed under the MIT License — see the LICENSE file for details.
