Pitch Narrative Builder
Turn an approved brand canon into an evidence-linked fundraising or sales deck narrative.
The skill declares least-privilege Tier-1 own-data inputs, treats pasted material as untrusted, labels unsupported proof as [needs source], and routes proposals through an authorized operation. However, rollback, sensitive-metric handling, and the complete permission boundary are not detailed here; publisher identity is also unverified, so points are deducted.
The workflow, stop conditions, NEEDS_INPUT behavior when no canon exists, audience confirmation, and abnormal-input handling are reasonably clear. However, it depends on several unprovided registries, scripts, and shared specifications, with no skill-specific tests or reproducible examples; under static calibration it remains below 10.
Triggers, audiences, input hints, expected outputs, and non-fit boundaries are clearly stated, with Chinese-language labeling and no connector prerequisite. Evidence is limited for host-by-host compatibility, depth of Chinese support, and mainland-China reachability, so points are deducted.
Frontmatter, version, license, Quick Start, contract, data sources, save rules, handoff, and next-skill guidance are well organized. Changelog evidence, maintenance ownership, update path, and availability of referenced files are not sufficiently demonstrated in the supplied material; publisher provenance is unknown, so points are deducted.
The skill specifies a five-beat narrative, slide mapping, claim IDs, audience-specific asks, and a pending-proof list, which plausibly yields a usable deck outline. No real output or execution evidence is supplied, and core results depend on external canon, story-bank, and claims-ledger state, so the score is conservative.
Claim-ledger binding, Measured/User-provided/Estimated labels, and event-based proposals provide some auditability. The supplied material contains no representative artifact, independent corroboration, or reproducible key-path record for this skill, so the score stays below the static ceiling.
- Before use, confirm that canon.md, the story bank, and the claims ledger exist and are version-consistent; without a canon the skill should stop with NEEDS_INPUT.
- Unapproved statistics, case outcomes, comparisons, or logos must remain [needs source] and must not be presented as verified pitch evidence.
- The skill may submit claims/narrative proposals or save results; verify that the host provides the authorized registry-events.py path and obtain explicit user confirmation before writes.
- No real execution artifact or skill-specific test is supplied, so the resulting deck outline requires human review.
What does this skill do, and when should you use it?
This skill derives a fundraising deck, sales pitch, or shared narrative from an existing narrative canon. It structures the story around five beats: problem, undeniable shift, product as hero, proof, and ask. Each beat is mapped to one or more slides, with proof beats tied to approved claim IDs or marked “[needs source].” It is a good fit for teams that already maintain a narrative canon, claims ledger, story bank, and usable traction data; without a canon, it stops rather than inventing one.
Reads the narrative canon, story bank, positioning truth set, read-only claims ledger, and user-provided traction or metrics; confirms whether the audience is sales, fundraising, or both; builds the problem→shift→product-as-hero→proof→ask arc; produces a one-beat-per-slide outline; identifies the proof unit and claim ID for each proof slide; marks unsupported evidence as “[needs source]” and routes it as a claims proposal; creates fundraising and sales-specific ask variants; and checks canon consistency, banned vocabulary, and AI-slop phrasing.
- A founder has an approved narrative canon and needs an investor-deck spine.
- A sales team needs a customer-facing pitch that matches the company’s approved message system.
- A marketing team wants the deck’s tagline, pillars, and proof wording to match the narrative canon and claims ledger.
- A team has traction, case studies, or comparisons that need clear measured, user-provided, or source-needed labeling.
What are this skill's strengths and limitations?
- Provides a clear five-beat arc and slide-level outline for deck planning.
- Requires proof beats to reference claim IDs and distinguishes supported from source-needed evidence.
- Supports sales, fundraising, or a shared body with audience-specific asks.
- Does not require connectors; the collection supports a Tier 1 workflow using supplied data.
- Requires an existing narrative canon and will not invent one when it is missing.
- Does not design finished slides or produce launch-window battle cards or rep talk tracks.
- Unsupported evidence remains a pending claims proposal rather than an approved marketing claim.
- The supplied material does not document an independent test suite or detailed platform-by-platform testing.
How do you install this skill?
In Claude Code, add the repository marketplace and install the collection: /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then run /plugin install aaron-marketing@aaron. Alternatively use npx skills add aaron-he-zhu/aaron-marketing-skills, or clone the repository with git clone https://github.com/aaron-he-zhu/aaron-marketing-skills. After installation, use narrative/land/pitch-narrative-builder/SKILL.md.
How do you use this skill?
Make sure memory/narrative-registry/canon.md exists and provide access to the story bank, claims ledger, and user-supplied metrics. Example prompts: Build our pitch deck narrative for [company] from the canon. Audience: investors. or Structure a sales pitch narrative for [product] — problem → shift → product-as-hero → proof → the ask. The skill produces narrative and slide beats only; finished visual design, launch-window battle cards, and sales talk tracks are out of scope.