"Labor omnia vincit" — Work conquers all
The complete multi-agent AI development team — orchestrate 6 specialized Claude agents from a single desktop app.
Operum (Latin: of works, from opus) automates your development, marketing, and community workflows with a team of AI agents that coordinate through a label-driven pipeline.
Get the latest Operum desktop app:
| Platform | Download |
|---|---|
| Linux | .deb · AppImage |
| Windows | .exe installer |
| macOS | .dmg |
graph LR
F([Founder]) --> PM[PM Agent]
PM --> A[Architect]
A -->|"ready-for-dev"| E[Engineer]
E -->|"needs-testing"| T[Tester]
T -->|"needs-review"| R[Review]
R --> D[Done]
T -.->|"FAIL"| E
Issues flow through the team automatically. Each agent owns specific pipeline stages, updates GitHub labels when done, and hands off to the next agent.
| Agent | Role | What It Does |
|---|---|---|
| PM | Orchestrator | Manages pipeline, delegates tasks, reports to founder |
| Architect | Technical Design | Reviews issues, provides implementation guidance |
| Engineer | Implementation | Writes code, creates PRs, runs tests |
| Tester | Quality Assurance | Tests PRs, approves or sends back for fixes |
| Marketing | Growth | SEO, content strategy, discoverability |
| Community | Support | Monitors channels, responds to users |
The only public repo with production-tested 6-agent Claude Code orchestration. Use these templates to build your own multi-agent AI system.
Define each agent's role, responsibilities, and step-by-step workflow.
| Agent | Role | Template |
|---|---|---|
| PM | Orchestrates the team, manages pipeline | pm.md |
| Architect | Reviews tasks, provides architectural guidance | architect.md |
| Engineer | Implements features, writes code, creates PRs | engineer.md |
| Tester | Tests PRs, reports results, approves for review | tester.md |
| Marketing | SEO, content strategy, growth initiatives | marketing.md |
| Community | Discord/Twitter monitoring, user support | community.md |
The coordination protocols that make multi-agent systems actually work.
| Template | Description |
|---|---|
| Pipeline | Label-driven issue flow: backlog → needs-architecture → ready-for-dev → in-progress → needs-testing → needs-review → done |
| IPC Protocol | File-based inter-agent communication with typed prefixes (DONE: / REQUEST: / ERROR:) |
| Branching | Git workflow with agent home branches (agent/{name}) and feature branch conventions |
| Release | Semantic versioning and CI-driven release management |
Bootstrap a new multi-agent project in minutes.
| File | Purpose |
|---|---|
CLAUDE.md.example |
Complete CLAUDE.md composing all templates |
project.toml |
Project metadata template |
goals.toml |
Goals tracking template |
View all templates and quick start guide →
No other public repository provides these patterns:
- Label-driven pipeline — Issues flow through defined stages with clear agent ownership
- File-based IPC — Agents coordinate via trigger/response files with typed prefixes
- Agent home branches — Each agent gets a persistent
agent/{name}branch for concurrent worktrees - Complete 6-role team — PM, Architect, Engineer, Tester, Marketing, Community with handoff protocols
- Event-driven coordination — Agents respond to events, with self-assessment as backup
Found an issue? Create a bug report
Please include:
- Operum version (shown in bottom-left of the app)
- Operating system
- Steps to reproduce
- Expected vs actual behavior
Have an idea? Request a feature
Questions or suggestions? Start a discussion
- Website: operum.ai
- Discord: Join our community (coming soon)
- Documentation: operum.ai/docs
| Term | Meaning | Context |
|---|---|---|
| Operum | Of works | Our name, from opus (work) |
| Opus magnum | Great work | What you'll build with Operum |
| Ex nihilo | From nothing | Start projects from scratch |
| Carpe diem | Seize the day | Ship faster with AI agents |
| Per aspera ad astra | Through hardships to the stars | The founder's journey |
Made with dedication by solo creators, for solo creators.