You need an RTL starting point
Good asks include "turn this architecture into RTL", "write a microarchitecture spec", or "create a front-end closure package for this block."
block-level-rtl-plan is the front-end flow that turns a high-level block brief into a Markdown
microarchitecture spec, synthesizable SystemVerilog, static audit results, and a downstream DV handoff package.
Good asks include "turn this architecture into RTL", "write a microarchitecture spec", or "create a front-end closure package for this block."
The flow starts with top_module, a short design_brief, and explicit performance, power, and area targets.
If power, performance, or area goals are missing, the flow records or asks for them instead of pretending it knows the closure target.
Turn the brief into deterministic REQ-NNN records, structured context, and explicit open questions.
Emit one Markdown spec with diagrams only when WaveDrom, Mermaid, or BlockDiag clarifies behavior.
Draft synthesizable SystemVerilog that traces back to the approved spec and requirement IDs.
Check for structural HDL issues, reset style problems, and synthesizability concerns before deeper analysis.
Classify risky crossings conservatively and carry blocking issues back into RTL revision when needed.
Check reset-domain crossings separately so reset safety is not hidden inside generic CDC review.
Estimate structurally hard paths and use them as advisory feedback unless the architecture clearly needs redesign.
Publish one final YAML handoff with requirements, spec path, RTL manifest, audit summary, unresolved items, and DV handoff metadata.