AI Tooling · Manual multi-agent coordination
Agent Desk
Coordinate model-specific prompts, dependencies, handovers and file-conflict checks for local AI-assisted development.
CodexClaude CodeWorkflowWhy it matters
The problem
Multi-agent work becomes difficult to audit when responsibilities, prompts, task dependencies and handovers are spread across tools.
How it works
The approach
A local-first CLI creates a structured task plan, prepares model-specific prompts, guards file scope and applies a deterministic quality gate.
What it includes
- Routed task plan and manual model selection
- Per-model copy-paste prompts
- File reservations and dependencies
- Structured handovers and quality gate
Try the tool
Quick start
Commands adapted from the project's documented README. Run from the repository root after cloning it; consult the full README for prerequisites and platform-specific setup.
git clone https://github.com/migmcc/AgentDesk.git
cd AgentDesk
python -m pip install -e ".[dev]"
agentdesk new "Demo Agent Desk"
agentdesk plan "Demo Agent Desk"
agentdesk status "Demo Agent Desk"Scope and trust
Important limitations
Agent Desk does not execute AI models, automatically push branches or create remote pull requests. It supports manual coordination; a full web UI is not currently implemented.