Package Exports
- @ikieaneh/opencode-kit
- @ikieaneh/opencode-kit/tui
Readme
opencode-kit
One plugin. Every project gets 18 agents, 40 skills, 16 slash commands, 5 MCPs, and contract-based orchestration, automatically.
Explore the docs »
Report Bug
·
Request Feature
Table of Contents
About
AI coding agents are inconsistent. Without structure, they skip conventions, ignore shared state, and bypass quality gates.
opencode-kit is an OpenCode plugin that replaces prose conventions with machine-readable enforcement. Install once — every project session gets agents, skills, MCPs, slash commands, and contract-based orchestration automatically.
Built with Node.js, OpenCode, lean-ctx, graphify, and GitNexus.
Getting Started
Prerequisites
- Node.js >= 18
- Git
- OpenCode installed and configured
node --version # >= 18
git --version # any recent versionInstallation
Step 1 — Install globally (one-time)
npm install -g @ikieaneh/opencode-kitStep 2 — Add to your project
{
"plugin": ["@ikieaneh/opencode-kit"]
}Add the snippet above to your project's opencode.json. Open your project in OpenCode — the plugin handles everything else.
Usage
After installation, verify everything loaded:
npx opencode-kit doctorExpected output:
✅ contract.json
✅ rules.json
✅ 18 agents provisioned
✅ 40 skills provisioned
✅ 7 rule files loaded
✅ All checks passedSlash Commands (16)
| Category | Commands |
|---|---|
| Health & Status | doctor, status, preflight, verify |
| Analytics & Scoring | analytics, score |
| Contract | contract-lint, diff, lock |
| Checkpoints | checkpoint, checkpoint-save |
| Project Tools | audit, init, update, version |
| ADRs | adr |
| Logs | logs |
All commands are prefixed with /opencode-kit: — e.g., /opencode-kit:doctor.
What You Get
Agents (18)
| Agent | Purpose |
|---|---|
orchestrator |
Delegates, validates, drives state machine |
planner |
Analyzes requests, traces impact, produces plans |
task-manager |
Breaks plans into tasks, implements each step |
code-reviewer |
Read-only code review — quality, security, performance |
explorer |
Fast codebase search across the entire project |
librarian |
Authoritative source for library docs and API references |
architect |
Strategic technical advisor for high-stakes decisions |
fixer |
Fast implementation for well-defined bounded tasks |
learner |
Post-execution learning — extracts lessons, persists knowledge |
observer |
System state monitor — read-only |
database-specialist |
Schema design, queries, migrations, optimization |
devops-agent |
CI/CD, deployment, infrastructure, automation |
documentation-agent |
Maintains README, API docs, inline documentation |
security-reviewer |
Vulnerability assessment and security best practices |
testing-specialist |
Unit tests, integration tests, test strategies |
council |
Multi-LLM consensus engine for high-stakes decisions |
designer |
UI/UX design, review, and implementation |
oracle |
Strategic technical advisor for architecture and debugging |
Skills (40)
| Category | Skills |
|---|---|
| Orchestration | orchestration-template, orchestration-workflow, dispatching-parallel-agents, executing-plans, subagent-driven-dev |
| Quality | test-driven-dev, systematic-debugging, verification-before-completion, quality-checks, receiving-code-review, requesting-code-review, simplify |
| Planning | writing-plans, brainstorming, using-git-worktrees |
| Web Research | firecrawl-search, firecrawl-scrape, firecrawl-deep-research, firecrawl-knowledge-base, firecrawl-knowledge-ingest, firecrawl-map, firecrawl-qa, firecrawl-workflows |
| Code Intelligence | gitnexus-exploring, codemap, token-optimize |
| Domain | database-design, sql-optimization, ci-cd, deployment, infrastructure, security-audit, testing-strategies |
| Learning | learner, using-superpowers, system-analyst, scoring-pipeline, qa-expert |
| Content | humanizer |
| Workflow | adr-generator |
MCPs (5)
| MCP | Purpose |
|---|---|
| lean-ctx | Token-compressed file/shell gateway (mandatory) |
| gitnexus | Code intelligence and impact analysis |
| context7 | Library documentation lookup |
| firecrawl | Web search, scraping, and interaction |
| github | GitHub API access |
Plus graphify (CLI tool, not MCP) for codebase graph intelligence.
.opencode-kit/ Storage
The plugin provisions agent templates, skills, rules, and contract files into .opencode-kit/ — all auto-managed, no manual setup needed.
Architecture
Inheritance Model
opencode-kit uses class inheritance: projects extend the base, not replace it.
{
"_meta": {
"extends": "opencode-kit",
"overrides": ["requirements.goal"],
"appends": ["scope.included"]
},
"requirements": {
"goal": "Project-specific goal (overrides base)"
}
}| Layer | Extend Via | Example |
|---|---|---|
| Contract | _meta.overrides / _meta.appends |
Override goal, append scope |
| Agents | _meta.append_skills |
Add project-specific skills |
| Rules | _meta.appends: ["rules"] |
Add custom rules |
| Skills | Add to .opencode/skills/ |
Create my-api-client/SKILL.md |
Merge rules:
- Scalars: project overrides base
- Arrays: concatenated + deduplicated
- Objects: deep merged (project wins)
- Excludes:
_meta.excludesremoves inherited items
See docs/inheritance-model.md for full architecture.
Roadmap
- Auto-provision agents, skills, rules
- Auto-config
opencode.json - ADR reports and session summaries
- Task complexity detection
- Graphify integration
- MCP availability checks
- Inheritance model with overrides and appends
- Scoring pipeline (PASS/RETRY/BLOCKED)
- Multi-language agent support
- Web-based dashboard
- VS Code extension
See ROADMAP.md for full details.
Contributing
See CONTRIBUTING.md for full guidelines.
Fork → Feature branch → Commit → PR. Follow the enforcement architecture. All rule enforcements must be tested.
License
Distributed under the MIT License. See LICENSE for more information.
Contact
RizkiRachman — GitHub
Project Link: https://github.com/RizkiRachman/opencode-kit