# Coordinate Multi-Agent Development Workflows

Complex projects need clear handoffs between planning, implementation, QA, and debugging agents. This skill defines a Serena Memory based workflow for coordinating specialized agents and monitoring progress.

## Install

```bash
npx skillstore add supercent-io/ohmg
```

## Metadata

- Status: approved
- Slug: supercent-io-ohmg
- Version: 1.0.0
- Author version: 1.0.0
- Skillstore revision: r1
- Version status: valid
- Tree hash: 460c989054ed8c4babb3800ecf802abf714e558c79d215bd2c625552a9586dbf
- Author: supercent-io
- GitHub username: supercent-io
- License: MIT
- Repository: https://github.com/supercent-io/skills-template/tree/main/.agent-skills/ohmg/
- Ref: f93e9bb0daca99badb6a7e574b97737155d57cb3
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/supercent-io-ohmg
- Manifest: https://skillstore.pages.dev/api/skills/supercent-io-ohmg/manifest

## Capabilities

- Explains when to use PM, frontend, backend, debug, and workflow guide agents.
- Defines Serena Memory as the shared state location for agent coordination.
- Lists setup, doctor, dashboard, bridge, update, and usage commands.
- Shows how to spawn a backend agent for a specific task.
- Documents per-agent CLI mapping for Gemini, Codex, and Claude.
- Describes terminal and web dashboards for workflow monitoring.

## Use Cases

- Coordinate Full Stack Delivery: Assign planning, frontend, backend, QA, and debugging work across specialized agents.
- Set Up Agent Workflows: Install the oh-my-ag workflow and verify local tools before starting coordinated work.
- Monitor Multi-Agent Sessions: Use Serena Memory and dashboards to follow status across active agent tasks.

## Prompt Templates

### Start Coordination

```
Use ohmg to coordinate this project: [goal]. Identify the agents needed and the first shared state entries.
```

### Plan Agent Roles

```
Break down [feature] into PM, frontend, backend, QA, and debug responsibilities using the ohmg workflow.
```

### Configure CLI Mapping

```
Create a per-agent CLI mapping for a project that uses Codex for backend, Claude for QA, and Gemini for frontend.
```

### Run Observability Review

```
Review my multi-agent session plan. Include Serena Memory checkpoints, dashboard monitoring, bridge needs, and security concerns.
```

## Limitations

- Requires external oh-my-ag, Bun, and uv tooling before workflows can run.
- Does not include implementation code for agents, dashboards, or Serena itself.
- Commands are examples and may change with the upstream package.
- Network bridge and dashboard setup need local security review.

## Best Practices

- Review every shell command before execution, especially installer, update, bridge, and dashboard commands.
- Keep Serena Memory entries structured so agents can resume work without guessing.
- Map each agent to a CLI that matches the task and your account limits.

## Anti Patterns

- Do not run unpinned package commands in sensitive repositories without approval.
- Do not spawn agents before defining shared goals, owners, and completion criteria.
- Do not expose bridges or dashboards outside localhost without authentication.

## Security Audit

- Audited at: 2026-07-07T05:00:41.103\+00:00
- Summary: The skill documents useful multi-agent coordination workflows, but several examples ask users to run unpinned external CLI commands. I found no prompt injection or credential exfiltration intent, but setup, update, dashboard, and bridge commands require explicit approval and safer pinning.

## Stats

- Views: 97
- Downloads: 10
- Favorites: 0
- Popularity score: 0