vibe-env-init

Scaffold any project to run opencode CLI inside a devcontainer, entirely from the terminal.

One command sets up .devcontainer/, .opencode/, and mise.toml.

Quick Start

bash <(curl -fsSL https://raw.githubusercontent.com/Miskamyasa/vibe-env-init/main/init.sh) my-project

The script will:

1. Create all configuration files with your project name
2. Skip any files that already exist (showing a diff so you can merge manually)

If no project name is given, the current directory name is used. The script normalizes it to a container-safe value (lowercase, a-z0-9_.-, separators collapsed to -).

If you need to test from a fork, override the template source:

VIBE_ENV_INIT_REPO="owner/repo" VIBE_ENV_INIT_BRANCH="branch" bash <(curl -fsSL https://raw.githubusercontent.com/Miskamyasa/vibe-env-init/main/init.sh) my-project

Running opencode in a container

After scaffolding, use the devcontainer CLI (installed via mise) to build and enter the container:

devcontainer up --workspace-folder .
devcontainer exec --workspace-folder . opencode

Wrapper script

To avoid typing the full commands every time, save this script as ~/.local/bin/oc (or any name you prefer):

#!/usr/bin/env bash
set -e

ROOT="$(git rev-parse --show-toplevel 2>/dev/null || pwd)"

# ensure container exists and is running
devcontainer up --workspace-folder "$ROOT" >/dev/null

exec devcontainer exec --workspace-folder "$ROOT" opencode

Then make it executable:

chmod +x ~/.local/bin/oc

Make sure ~/.local/bin is in your PATH (add export PATH="$HOME/.local/bin:$PATH" to your shell profile if needed).

Now you can run oc from any scaffolded project directory to start opencode inside its container.

Session Mode

The scaffolded environment uses the latest opencode with SQLite-based sessions. Each project gets its own isolated opencode data directory at ~/.cache/containers/local-share-opencode/<project>. Auth credentials are mounted read-only from the host.

What Gets Created

your-project/
  .devcontainer/
    devcontainer.json     # Container configuration
    Dockerfile            # Container image definition
  .opencode/
    opencode.json         # opencode CLI configuration
    tui.json              # Terminal UI theme and plugin configuration
    agents/
      build.html          # Direct-use implementation agent
      explore.html        # Fast codebase exploration sub-agent
      general.html        # Scoped implementation sub-agent
      investigate.html    # Read-only investigation agent (default)
      lead.html           # Orchestration agent for commands
      plan.html           # Planning agent
      review.html         # Code review sub-agent
    commands/
      execute.html        # Implementation execution command
      review.html         # Review orchestration command
  mise.toml               # Tool versions (node, npm, devcontainer CLI, opencode, gh)

Usage Example: Execute and Review Workflow

The scaffolded .opencode/ directory provides a two-phase workflow for implementing features. Each phase uses a dedicated slash command and specialized AI agents. Use the plan agent first to produce an execution-ready plan, then run /execute and /review.

Step 1: Initialize the devcontainer

bash <(curl -fsSL https://raw.githubusercontent.com/Miskamyasa/vibe-env-init/main/init.sh) my-project

Then start the container:

devcontainer up --workspace-folder .
devcontainer exec --workspace-folder . opencode

Or use the oc wrapper if you set it up (see Wrapper script).

Step 2: Plan (manual)

Switch to the plan agent and describe what you want to build:

add a homepage benefits section

The plan agent (read-only) will:

• Read AGENTS.md and any referenced files for project conventions
• Investigate the codebase for existing patterns and architecture
• Produce a dependency-ordered list of implementation steps with acceptance criteria

Wait for the plan to complete. Review the output — it will contain numbered steps (S1, S2, ...) with scope, dependencies, and acceptance criteria for each. If required context is missing, the agent reports "Missing Context" and halts instead of guessing.

Step 3: Execute (/execute)

Once you are satisfied with the plan, run:

/execute

The /execute command uses the lead agent to:

• Walk through each planned step in dependency order
• Spawn general sub-agents for each step with implementation-only prompts
• Handle failures by inserting remediation steps and retrying
• Produce an execution summary with status per step and list of modified files

Wait for execution to finish. The agent will report which steps succeeded, failed, or were blocked.

Step 4: Review (/review)

After execution completes, run:

/review

The /review command uses the lead agent to spawn read-only review sub-agents that:

• Verifies each planned step was implemented correctly (plan fidelity)
• Checks for regressions at API boundaries, state transitions, and error paths
• Validates acceptance criteria coverage
• Flags maintainability issues in touched areas

If the review finds critical issues, the agent will propose fix steps and wait for approval. Once the review passes, you get:

• A structured review report with findings and verdict
• A proposed commit message (title + body with list of changes)

Step 5: Commit

Copy the proposed commit message from the review output and commit the changes yourself:

git add -A
git commit -m "feat: add homepage benefits section

- Add BenefitsSection component with responsive grid layout
- Create benefits data module with icon mappings
- Integrate section into homepage below hero block
- Add unit tests for BenefitsSection rendering"

Review the diff before committing — the proposed message is a suggestion, not a guarantee.

Agents Reference

Agent	Model	Mode	Purpose
investigate	default	primary (default)	Read-only codebase investigation, mapping structure, tracing dependencies
plan	default	primary	Produces execution-ready implementation plans (read-only)
build	default	primary	Direct-use agent for implementing changes end-to-end
lead	default	primary	Orchestrates /execute and /review commands, delegates to sub-agents
general	opencode-go/deepseek-v4-flash	subagent	Scoped implementation tasks with minimal changes
explore	opencode-go/deepseek-v4-flash	subagent	Fast, broad codebase exploration for investigation
review	openai/gpt-5.6-sol	subagent	Read-only code review with severity-rated findings

Commands Reference

Command	Agent	Description
/execute	lead → general	Execute the plan step-by-step via implementation sub-agents
/review	lead → review	Run independent code review and produce commit message

Tip: The investigate agent is the default mode. When you open opencode, you can ask questions about the codebase and it will explore in read-only mode without making any changes. Switch to the plan agent when you are ready to design an implementation, then run /execute and /review to implement and verify it.

Conflict Handling

If a file already exists, the script will:

• Skip the file (never overwrites)
• Show a unified diff between your file and the template
• Print the template URL so you can review and merge changes manually

Prerequisites

• macOS or Linux
• curl or wget
• Docker (or a compatible container runtime)
• mise for tool management (installs devcontainer CLI, opencode, etc.)

License

MIT