creative-strategist — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited creative-strategist (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.
Lock a brand's visual direction before any assets are generated, and encode it as a short, reusable style block — a fragment of prompt language that gets pasted into every downstream generation prompt. Prime rule: consistency beats beauty. Ten assets in one coherent style build a brand; ten gorgeous assets in ten styles build noise. The brief exists so that image prompts written weeks apart still produce siblings.
Use at the start of any creative engagement where more than one asset will be made, when generated assets "don't look like the same brand", or when the user can't articulate their aesthetic. Hand off instead when:
straight to skills/creative/image-generation or the specialist skills.
skills/creative/orchestrator or the specific asset skill.
skills/product-strategy/foundation-sprint first; its customer/differentiators feed this brief.
Ask in one batch, only what's missing:
use? (Ask for hex codes or links if they exist.)
Infer, don't ask: industry conventions (fintech skews trustworthy/clean, D2C lifestyle skews warm/photographic); primary destinations from context (an e-commerce user needs product-photo direction more than illustration direction). Don't stall: with brand + audience known, propose the rest as labeled assumptions and proceed.
spectrums: photorealistic ↔ illustrative, minimal ↔ maximal, warm ↔ cool, playful ↔ serious, mass ↔ premium. Note where competitors sit — matching them is a choice, not a default; flag one deliberate point of visual contrast.
vibe, a 3–5 color palette (with hex codes), lighting + composition defaults, and one sentence on why it fits the audience. Make them genuinely different directions (e.g. "Warm Editorial" vs "Clinical Minimal" vs "Bold Flat Illustration"), not shades of one idea. The user picks one; you may recommend, with a one-line reason.
format below. No field may read "flexible" or "depends" — a value the user can overrule later beats a blank.
prompt language, e.g.: photorealistic, warm golden-hour lighting, rule-of-thirds composition, warm gold (#C9A227) and cream (#F5EFE0) palette, approachable premium mood, highly detailed, professional photography Test it mentally against three different asset types (product shot, social graphic, illustration) — if it only works for one, generalize the wording and move medium-specific details into the per-medium adaptations.
(e.g. "never neon colors", "no stock-photo handshake clichés", "no more than two typefaces") and the rule for when secondary style variants may be used.
FAL_API_KEY is set,generate one probe image per finalist direction so the user picks from pixels, not adjectives: python docs/creative_cli.py custom --category style-tests --name <direction-slug> --prompt "<representative subject>, <style block>" --resolution 1K (Full automation reference: skills/creative/image-generation/references/automation.md.)
Deliver exactly this brief (markdown, concise):
# Creative Direction Brief — [brand]
## Chosen direction: [name] (rejected: [alt 1], [alt 2] — one-line why)
## Aesthetic profile
| Field | Value |
| Primary style | photorealistic / illustrative / minimal / abstract / mixed: ... |
| Color palette | [name #hex] x 3–5, with roles (primary / secondary / accent) |
| Mood | [3 adjectives] |
| Lighting default | [e.g. soft studio, natural golden hour, dramatic rim] |
| Composition default | [e.g. centered + negative space, rule of thirds] |
| Detail level | minimal / moderate / high |
| Audience | [who this must appeal to] |
| Deliberate contrast | [the one way we look different from the category] |
## Style block (paste into every image prompt)
> [1–3 lines of literal prompt language]
## Per-medium adaptations
- Product photography: [what changes — background, lighting]
- Social graphics: [what changes — contrast, text space]
- Brand assets / illustration: [what changes — flatness, line weight]
## Prohibitions
1. Never ...
2. Never ...
## Consistency rule
Every prompt for this brand includes the style block verbatim; deviations are
proposed to the user first, never improvised.Hard don'ts: never deliver a brief with TBD fields; never let the user pick "all three directions"; never restyle mid-project without updating this brief first.
skills/product-strategy/foundation-sprint → upstream; its target customer anddifferentiators become this brief's audience and deliberate contrast.
skills/creative/image-generation → consumes the style block in everyprompt spec's style/lighting/mood fields.
skills/creative/product-photography, skills/creative/social-graphics,skills/creative/brand-asset → consume the style block plus their per-medium adaptation line.
skills/creative/orchestrator → runs this skill first on any multi-asset pathwhere no brief exists; the brief is the artifact passed forward.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.