venture-pitch-builder — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited venture-pitch-builder (Agent Skill) and scored it 96/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 1 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 1 flagged
The text {match} tells the agent to skip the normal "ask the user first" gate. Used adversarially it removes the human-in-the-loop check before destructive or sensitive actions, turning a normally-gated agent into a fire-and-forget executor.
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
This skill produces two linked things from a startup idea: (1) a clear, research-backed strategy grounded in the Bling Capital Diligence Playbook, and (2) a living Vite + React pitch/diligence deck — a multi-tab web app whose numbers are computed from one model, so the founder can tune an assumption and watch the whole deck update.
The methodology is in references/bling-playbook.md — read it before mapping research into the model. A complete, known-good worked example is in assets/reference-app/ — that is the canonical structure to copy and re-theme, not rebuild from scratch.
Work conversationally throughout. This is a collaboration about the founder's business, not a one-shot generator. Float the strategy, let them push back, then build.
The idea as stated is almost always underspecified, and the research is only as good as the framing. Ask 2-3 short shaping questions with the AskUserQuestion tool before researching. Pick the questions that matter most for this idea; good candidates:
Keep it light — offer concrete options, let them free-text. Weave the answers into the research query and the model. If the user already gave rich detail, ask fewer questions (or just confirm scope).
deep-research skillInvoke the `deep-research` skill (via the Skill tool) with a single, well-framed query built from the idea + interview answers. Frame it around the facts the Bling exercises need, so the report comes back diligence-ready. Ask it to find, with sources:
plus top-down industry spend for context only. Prefer counts of customers/venues/companies over a single TAM dollar figure.
roughly how many customers per segment.
whether budget already exists.
(this anchors the "10x better" and "why we win" claims).
If the deep-research skill is unavailable for any reason, fall back to running web searches (WebSearch / web_fetch) yourself against the same checklist — but prefer delegating.
While research runs, you can draft the strategy skeleton from the interview answers.
Read references/bling-playbook.md, then turn the research into the four exercises. Do the math explicitly before touching code — these become the constants in the app's data/ files:
step-by-step today flow and tomorrow flow; a quantified 10x.
price/ACV and gross margin per segment.
X customers at $Y = K% of the pool to reach $100M gross profit; then the $500M story.
right-sized to the cash trough.
Share this strategy with the founder in chat (concise prose, the key numbers and the mad-libs) and invite corrections before building. Their domain knowledge will sharpen the assumptions.
Copy the reference app into the user's working folder under a project-appropriate name, then re-theme and refill it — don't regenerate the scaffolding. The architecture, build config, responsive CSS, shared components, and the computed-model pattern all transfer directly; only brand, data, and copy change.
cp -r <skill>/assets/reference-app <workspace>/<company>-pitch
cd <workspace>/<company>-pitchThen edit, in this order:
the hex values). Match the real brand if a site exists.
CORE_MARKET /EXPANDED_MARKET), per-customer economics, the SEGMENTS definitions, penetration, and the $100M/$500M path functions. Keep the exported function names (buildPath, customersForGP, customerGrossProfit, GROSS_MARGIN, …) so the tabs keep working; change the numbers and segment definitions. The example is a B2B SaaS segmented by company size — re-shape it for a marketplace (segment by geography/tier), transactional, or usage model as needed.
burn. Re-derive buildProjections, fundingNeed, and revenueMix for the business model at hand — the example uses seats/ACV per segment, but the structure is a guide, not a straitjacket.
[n], with the real sources from research,plus the methodology note.
default; rename, drop, or add tabs to fit the business (e.g. a "Team" tab).
strategy. The tab→exercise map is in the reference table in bling-playbook.md. Keep the visual components (H2, InfoCard, FlowStep, SectionNote, Tag, StatCard) from components/Shared.jsx; they're generic.
Principles that make the deck good (the same ones the reference app embodies):
data/ so one assumptionchange ripples everywhere. Resist pasting literal figures into JSX.
reference does, and cite real sources. Investors trust clear, evolving plans over polished fiction.
cd <workspace>/<company>-pitch
npm install
npm run build # must succeed — fix any errors before presenting
npm run dev # for the founder to view locally (print the localhost URL)A successful npm run build is the verification gate: it catches missing imports, renamed exports, and JSX errors. Then tell the founder how to run it (npm run dev), and present the key files / the strategy summary. Because the model is live, invite them to tune the assumptions in data/financials.js and data/market.js — and offer to make changes with them. Keep iterating conversationally: this deck is a living document.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.