Overview
What Pegboard does, and where to go next.
Every coding agent stores its MCPs and skills in a different place and format. Keeping them in sync means translating by hand, or asking an agent to do it and hoping it gets every format right.
Pegboard replaces that with one file. You declare what your agents should have in pegboard.yaml, and pegboard writes each agent’s native config.
pegboard plan # what would change; writes nothing
pegboard apply # applies it, backing up every file
pegboard drift # did anyone edit managed files by hand?
If you’ve used Terraform, the workflow will feel familiar. Translating config is deterministic, tested code, so the result is the same every time, and it costs no tokens.
Why the name
A pegboard is the perforated workshop board you hang tools on. On a shadow board, each tool has a painted silhouette, so you can see at a glance what’s missing or out of place.
| In the workshop | In Pegboard |
|---|---|
| The tools | MCPs and skills |
| The painted silhouettes | pegboard.yaml, your desired state |
| Spotting an empty silhouette | pegboard drift |
| Putting each tool back | pegboard apply |
Where to go next
Current status
Pegboard is an early MVP. MCPs, skills, and AGENTS.md work across all nine supported agents.
These are planned and don’t work yet:
includefor remote and shared stacks, which is currently ignored with a warningop://andkeychain:secrets- A TUI, and a
pegboard mcpserver