Main flow today

Build one block-level DV plan from design intent, RTL, and optional risk reports.

This is the main multi-step flow implemented in the repo right now. It is for planning verification work for a block or small IP. It does not generate runnable UVM code.

View flow source See the steps

Overview Start Here Skills Rules Block DV Flow

Use it when

You need a verification plan

Good asks include "build a DV plan", "derive DV objectives", or "plan tests, assertions, and coverage."

Main inputs

Intent + RTL

The flow starts with `top_module`, `rtl_files[]`, and a short `design_intent` brief.

Optional inputs

CDC or timing risks

Existing CDC or timing reports can be imported as risks, not as proof of full signoff.

Step by step

The flow runs six skills in a fixed order.

Derive DV objectives

Turn the design brief and visible RTL behavior into stable objective IDs.

Extract verification surface

Map clocks, resets, interfaces, key state, observability points, and unresolved gaps.

Plan UVM tests

Choose the smallest justified environment and a prioritized test list.

Plan assertions

Create conservative SVA candidates that match visible block behavior.

Plan coverage

Create objective-linked coverpoints, crosses, and exclusions.

Assemble final plan

Merge all intermediate artifacts and import optional CDC or timing reports as risk inputs.

Output

One final YAML plan

The end result is one structured DV plan with objectives, interfaces, environment notes, tests, assertions, coverage, risks, unresolved items, and summary counts.

design -> objectives -> interfaces -> env -> tests -> assertions -> coverage -> risks -> unresolved -> summary

Not for

Know the limits

It does not generate runnable UVM class code.
It does not debug regressions.
It does not plan SoC-level verification architecture.
It does not claim signoff completeness.

Example

Timer counter flow example

Intent input

Short brief

The timer counter example says the block is CSR-controlled, can load a count, raises `irq` at terminal count, and clears to an idle state on reset.

Flow result

Structured plan

The assembled plan keeps tests, assertions, and coverage tied to objective IDs and imports timing risk without inventing CDC issues.

Rules used across the flow

The flow applies evidence grounding, output discipline, objective traceability, UVM component selection, assertion classification, coverage taxonomy, and stimulus prioritization.

Read the rules Review the skills Open upstream FLOW.md