Instructions file
Instructions for sean-galloway/RTLDesignSherpa, covering claude code guide for rtl design sherpa, the handbook (repo memory) - read this first, repository philosophy, core principles and 📖 global requirements reference.
Agent
The five repo-resident agent roles, what each may touch, and the stub that makes it discoverable. Roles are scoped to this repo, never to the machine.
Agent
The role that writes testbenches. Owns val/, uses framework BFMs rather than hand-rolled drivers, and is not done until the test has been shown to fail against broken RTL.
Agent
The role that writes and runs formal properties. Owns formal/, and treats a vacuous pass as a failure - a property that cannot fail has proven nothing.
Agent
One working tree, several agents - the four ways uncommitted state crosses agent boundaries, and the staged-set check that actually holds.
Agent
The role that runs regressions and triages failures. Produces verdicts and quarantines, never fixes - and never reports a verdict that a clean rebuild has not reproduced.
Agent
The role that writes RTL. Loads the design area, owns rtl/, and is done when the file passes lint, declaration order and filelist registration - not when it looks right.
Agent
The read-only role that judges RTL it did not write. Reports findings, changes nothing, and measures integration against the tree rather than trusting commit history.