opencode-onboard logo opencode-onboard

Prepare any codebase for AI

One interactive wizard that wires OpenCode, OpenSpec, codegraph, and agentmemory into a multi-agent development workflow powered by native parallel subagents.

$ npx opencode-onboard@latest
npm version npm downloads license node version

GitHub · Azure DevOps · Jira · GitLab · browser-based backlogs, or any combination.

your-project: opencode-onboard

Agents shouldn't improvise

Most codebases have no AGENTS.md, no architecture docs agents can read, and no defined workflow for picking up tasks. Agents end up improvising, producing inconsistent results. opencode-onboard fixes that in a single interactive wizard. It installs an agent team, platform skills, slash commands, and structured change management. Everything agents need to plan, implement, and ship.

What it wires together

OpenCode

The agent runtime. Native parallel subagent waves, no external plugins, no git worktrees.

OpenSpec

Structured change management: proposals, specs, and tasks for every piece of work.

codegraph

Code intelligence and indexing so agents understand your codebase before touching it.

agentmemory

Shared context across agent sessions, decisions persist beyond a single conversation.

The 10-step wizard

Run it once. Progress stays visible the whole way through.

  1. Source scope

    Choose current repo or sibling source roots for code analysis.

  2. Clean AI files

    Detects existing AGENTS.md, .cursorrules, CLAUDE.md and removes them, your .agents/skills/ are preserved.

  3. Choose platform

    Backlog (GitHub / Azure DevOps / Jira / Browser / None) + repo (GitHub / Azure DevOps / GitLab / None). Mix freely.

  4. Check platform CLI

    Verifies gh, az + azure-devops, acli, or glab as needed.

  5. Copy scaffolding

    Agents, built-in skills, bootstrap docs, source-roots metadata, then runs npx skills.

  6. Init OpenSpec

    Runs npx @fission-ai/openspec init silently for structured change management.

  7. Choose models

    Live list from models.dev, pick plan / build / fast models with canonical cost indicators.

  8. Token optimization

    Optional checklist: RTK, opencode-quota, caveman, and dynamic rule injection into AGENTS.md.

  9. Browser plugin

    Installs @different-ai/opencode-browser for agent browser automation.

  10. Write metadata

    Saves .opencode/opencode-onboard.json so reruns and teammates reuse your setup.

Then open OpenCode in your project and type /ob-init, brownfield projects get ARCHITECTURE.md and DESIGN.md generated from the actual codebase.

Slash commands, installed and ready

Custom commands land in .opencode/commands/ and are available directly in OpenCode.

/ob-help

Show all commands and when to use each one. Start here if you're unsure.

/ob-init

Initialize the project, greenfield or brownfield, and activate the agent team.

/ob-explore

Think through an idea or investigate a problem before committing to a plan.

/ob-propose

Parse an issue URL or idea into a structured plan: proposal, specs, tasks.

/ob-apply

Implement tasks via parallel subagent waves with the native task tool.

/ob-pullrequest

Create a PR for the current branch, or read and classify review comments.

/ob-archive

Archive a completed OpenSpec change.

/ob-main

Quick direct implementation, no OpenSpec, no waves, no PR. Just do it.

/ob-autopilot

Autonomous pipeline: branch → propose → apply → archive. One commit per phase.

/ob-create-engineer

Create a custom specialist engineer with skills auto-installed from skills.sh.

/ob-create-architecture

Generate or regenerate ARCHITECTURE.md from the codebase.

/ob-create-design

Generate or regenerate DESIGN.md from the design system.

/ob-create-project-guardrails

Generate a project-guardrails skill from ARCHITECTURE.md and project config.

/ob-set-model

Set the model for a tier (plan, build, or fast) per team or per user.

The pipeline

Give the lead agent a work item URL and execution flows through a defined pipeline.

1

Parse work item

The platform userstory skill turns a GitHub / Azure DevOps / Jira / browser URL into a structured work item.

2

Propose

/ob-propose produces proposal.md, specs, and tasks.md, enriched with agent and model assignments.

3

Confirm

You review the plan before any implementation starts.

4

Parallel subagent waves

Engineers spawn in parallel, each carries its own model tier, implements its tasks, and returns. The lead commits each group. Watch it live in the Subagents panel.

5

Verify

Tests, build, and lint run according to task scope.

6

Ship

Commit → push → PR → review feedback loop, handled by the lead.

Agents, skills, and model tiers

Agents: how to work

Universal personas: a lead orchestrator plus ability-driven engineers. Add project specialists with /ob-create-engineer.

Skills: what to know

Platform knowledge and project rules, auto-detected and auto-loaded. Drop a SKILL.md in .agents/skills/ and agents pick it up.

Models: plan / build / fast

Three tiers picked from 3000+ live models: strong reasoning for the lead, capable for engineers, fast and cheap for helpers.

See it run

opencode-onboard wizard demo

Get started

$ npx opencode-onboard@latest

Requires Node.js 18+ · Joining an already-onboarded project? Run npx opencode-onboard join

Star on GitHub Read the full docs