elevator-pitch — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited elevator-pitch (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
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.
You are producing the shortest defensible pitch for a project. Two formats:
The constraint is the discipline. If you can't pitch the project in 150 words you don't understand it yet.
Both rules are enforced by validators at the end of generation. If either fails, revise before surfacing.
A compressed answer to four questions, in this order:
That's the four-beat shape. Adapt the verbs and nouns per audience.
If you find yourself writing "Imagine a world where..." or "Our mission is to..." — back up. Those aren't pitches; they're avoidance.
generalist (default)Anyone — a friend at a party, someone on a plane, a non-technical curious person. Plain words only. Stage information matters; people want to know if this is real or aspirational.
investorSomeone evaluating commercial potential.
technicalA developer who'll evaluate the technology choice.
partnerA potential integration / partnership / collaboration target.
In parallel:
README.md at repo root (primary source for the pitch)package.json / pyproject.toml / etc.) for description + versiondocs/vision.md if present (target users, problem statement, success signals)docs/roadmap.md if present (stage signal)git log --oneline -10 (stage / activity signal)If --audience wasn't passed, ask. If --output wasn't passed, offer:
PITCH.md at repo rootFour beats. Write each beat to a single short sentence.
Example shape (rad-explain itself, hypothetical):
For developers using Claude Code who need to explain their project to others —
rad-explain generates project narratives, elevator pitches, and grounded README
audits from your repo source. Other AI tools will puff your project; rad-explain
runs a grounding validator that flags any claim it can't trace back to your
code or docs. v0.1.0 — five skills + two validators, in active use in this
marketplace.That's ~75 words, ~30 seconds spoken, four beats: who-for / what-it-does / vs-alternative / stage.
# Write to /tmp/pitch-draft.md
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/check-grounding.py /tmp/pitch-draft.md --repo <repo-root> --json
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/check-overpromise.py /tmp/pitch-draft.md --jsonCommon overpromise mistakes in elevator pitches:
If output target is stdout, print the pitch. If file, write and surface a short summary:
Elevator pitch written to {path}.
Audience: {audience}
Word count: {N} (target: 120-180 for written, 70-90 for spoken)
Validators: grounding {pass|N flagged}, overpromise {pass|N flagged}
Read aloud once. If it doesn't sound like something you'd actually say, revise.draft-pitch in this plugin.narrate-project in this plugin.scripts/ dir.~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.