Skip to main content

core-agents Bundle

The core-agents bundle scaffolds a complete AI agent team into your repository, ready for spec-to-ship workflows.

What gets installed

AgentRole
ag-architectTurns feature requests into structured specs, context maps, and ADRs
ag-coderImplements backend code from an approved spec
ag-ui-developerImplements UI code from an approved spec
ag-testerWrites and runs tests, enforces TDD

A workflow.md file is also installed (e.g. .github/agents/workflow.md) with the full spec-to-ship sequence.

Install

# Default stacks: dotnet backend, react frontend
ag-skill add core-agents

# Choose a different backend
ag-skill add core-agents --backend python --frontend react
ag-skill add core-agents --backend java --frontend react
ag-skill add core-agents --backend go --frontend react

Stack options

FlagValues
--backenddotnet (default), python, java, go
--frontendreact (default)

See Stack Variants for full details.

How to use the agent team

  1. Open your AI agent (e.g. GitHub Copilot Chat)
  2. Select ag-architect and describe your feature
  3. Architect produces a spec, context map, ADRs, and a summary of artifact file paths
  4. Approve explicitly when prompted (for example, reply approved) to allow handoff
  5. Hand the approved spec to ag-coder and ag-ui-developer
  6. ag-tester writes and runs tests against the implementation

See the Workflow Usage Guide for step-by-step instructions.