Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/chipweaver/veripower/design-flownpx skills add chipweaver/veripower --skill design-flowgit clone --depth 1 https://github.com/chipweaver/veripowerWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/chipweaver/veripower/design-flow)<a href="https://agentmods.dev/skills/chipweaver/veripower/design-flow"><img src="https://agentmods.dev/badge/skills/chipweaver/veripower/design-flow.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00031 | $0.02059 |
| Opus 5 | $0.00015 | $0.01030 |
| Sonnet 5 | $0.00006 | $0.00412 |
| Haiku 4.5 | $0.00003 | $0.00206 |
Grade A, and why
design-flow scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 149 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Design Flow Orchestrator
You are the Orchestrator. Every turn you run one deterministic step against the
kernel tool python3 <skill>/../../framework/scripts/kernel.py (written kernel.py below).
<skill> is this skill's own base directory, named on the first line of this file.
loop:
a = kernel.py decide --module {module} [--wake <rule>:<run>] [--closing]
execute(a) # a.action ∈ {DISPATCH, REAP, YIELD, DONE, ESCALATE}
# non-blocking for `execution: task`
if a.action in {YIELD, DONE, ESCALATE}: end turn
{module} is the path to the module's directory — the one holding events.jsonl and
brainstorm.md, with every stage's tree beneath it. Pass the same one every call; an
absolute path works from any working directory.
kernel.py is the sole writer of {module}/events.jsonl and the sole decider.
decide reads on-disk state and returns exactly one action as a JSON object on exit 0. All
routing lives inside it — you never re-derive the next stage yourself, and you carry nothing
between turns: what to build is derived from the log every call, so a compaction or a crash
costs you nothing but the turn.
Every verb prints a JSON envelope. An ok: false is a contract signal, not an obstacle:
hand the error to the user and stop that line of work. A non-zero exit means the
call could not be made sense of at all (today: no module directory at the resolved path,
which is a wrong working directory) — follow the printed error, don't debug the script.
kernel.py status --module {module} prints the per-stage projection plus signed_off. It
is a read-only query for the user, outside the loop.
Iron Rule
- Do not run EDA tools (make / vcs / dc_shell / pt_shell / spyglass) yourself — that is the stage subagent's job.
- Do not hand-edit
events.jsonlor any stageresult.json/ artifact. A main-thread write to either is an isolation violation. - Scripts are black boxes — never Read their source. Invoke
kernel.pyper this skill's documented command lines (flags via<verb> --help). Sole exception: debugging a suspected bug in the script itself.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 149 lines · 31 tokens per session scan A c431c99e1ef2
design-flow is a skill published in the GitHub repository chipweaver/veripower (4 stars, last pushed 5d ago), licensed MIT. It adds 31 tokens to every session and 2,059 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
vdesign
Generate a design proposal, documentation, and RTL backbone from a requirement file.
varch
Decompose an IP requirement into submodules, define interfaces, and generate an architecture document plus per-module requirement files and a top-level skeleton.
vfill
Implement Verilog code from an approved design proposal, then lint and simulate.
vflow
Scan project state to determine phase completion, write flowstatus.md, and orchestrate the full RTL design flow from architecture to documentation.
vsynth
Synthesize a Verilog module or full IP with Vivado and report resource utilization, Fmax, and power.
vexplain
Generate documentation for any RTL module from Verilog source files.