AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Brief Graphic

skill-hungv47-meta-skills-brief-graphic · by hungv47

Produce a graphic-design brief for a single visual asset — social posts, thumbnails, banners, OOH, OG cards, hero illustrations. Pulls brand tokens, generates concepts, writes platform-aware specs + image-gen prompt or designer-handoff spec. Does NOT render — rendering is downstream. Not for brand identity (use create-brand), whole-page redesigns (use brief-landing-page), or the copy that goes IN…

No reviews yet
0 installs
36 views
0.0% view→install

Install

$ agentstack add skill-hungv47-meta-skills-brief-graphic

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-hungv47-meta-skills-brief-graphic)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
2mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Brief Graphic? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

Design Brief — Orchestrator

Produce a graphic-design brief for a single visual asset — social posts, thumbnails, banners, OG cards, hero illustrations, OOH, ad creative. Emits an image-gen prompt or designer-handoff spec; does NOT render (not for user flow / wireframe — use map-user-flow). Capability metadata: [routing.yaml](routing.yaml). Agents + 4 downstream routes + 8-dim rubric: [references/agent-manifest.md](references/agent-manifest.md). Methodology: [references/playbook.md](references/playbook.md).

Core question: Could a designer or image-gen tool execute this asset on-brand and on-platform without follow-up questions?

Critical Gates — load first

  • Do NOT render. Brief only — spec + reference direction + image-gen prompt. Rendering happens downstream.
  • Do NOT proceed without brand anchors. Missing brand/BRAND.md or brand/DESIGN.mdNEEDS_CONTEXT, recommend create-brand. Hard gate fires BEFORE Cold Start (survives --fast).
  • Do NOT invent tokens, fonts, or motion specs. Every visual decision traces to DESIGN.md. Gap → flag, don't guess.
  • Reference realized surfaces, not just tokens. Before writing reference direction, pull ≥1 realized surface (live page screenshot · shipped HTML · approved exploration · existing exemplar) and design the concept against it — tokens are the floor, the realized surface is the taste. No surface available → record the explicit token-only fallback line. Silent token-only design fails the rubric. Contract: [references/_shared/realized-surface-grounding.md](references/_shared/realized-surface-grounding.md).
  • Do NOT use stock-AI defaults. No purple-blue gradients, centered-on-white, faux-3D bevels, glassmorphism — unless DESIGN.md specifies. 13-pattern detector scores explicitly.
  • Do NOT skip the brief approval gate. Brief is a candidate, not a delivery — user reviews before rendering.
  • Platform spec mandatory. Every brief includes aspect ratio, safe zones, mobile readability, file format, file-size cap.

Quality Gate — 8 dimensions + AI-aesthetic check

Critic verifies. Full rubric + per-route scoring shifts + 13-pattern Generic-AI-Aesthetic Detector: [references/visual-rubric.md](references/visual-rubric.md) and [references/agent-manifest.md](references/agent-manifest.md) § 8-Dim Visual Rubric. Max 2 rewrite cycles per verdict.

  • [ ] Brand fidelity + sacred-element respect — palette/type/motion trace to DESIGN.md; no proposed change to logo, anchor color, tagline unless brief authorizes
  • [ ] Hierarchy + composition + typography + contrast — 1-sec focal point, intentional white space, pairing per DESIGN.md, WCAG AA (≥4.5:1 normal, ≥3:1 large)
  • [ ] Platform fit — dimensions, safe zones, crop behavior, mobile thumb-stop readability, file format/size match the platform module
  • [ ] CTA clarity (if applicable) + Generic-AI-Aesthetic (0-7 clean / 8-15 DONEWITHCONCERNS / 16+ auto-FAIL)
  • [ ] Downstream handoff complete — image-gen prompt specific (lens/lighting/mood/era/composition/color cast), or designer spec covers placement + tokens, or vector spec specifies layout grid

Before Starting

Apply [references/_shared/before-starting-check.md](references/_shared/before-starting-check.md).

| Artifact | Source | Required? | |---|---|---| | brand/BRAND.md | create-brand | REQUIRED (hard gate) — voice, archetype, sacred elements | | brand/DESIGN.md | create-brand | REQUIRED (hard gate) — palette, type, surface, motion | | brand/CREATIVE-DIRECTION.md | create-brand | Recommended — house art direction (mood, light, framing, motion); design against it, not just tokens | | brand/ASSETS.md | create-brand B | Optional — auto-fill dimensions + checkbox tick | | realized surface (live URL · shipped HTML · explorations/* · exemplar asset) | the live brand | Recommended — the executed taste to design against; see realized-surface-grounding | | docs/forsvn/artifacts/marketing/brief-landing-page/[slug]/asset-slots/[slot-id].md | brief-landing-page | Optional — LP slot spec | | docs/forsvn/artifacts/marketing/content/[slug].copy.md | write-copy | Optional — copy that renders IN the asset | | docs/forsvn/artifacts/marketing/campaign-plan.md | plan-campaign | Optional — campaign context + ## Creative Direction (per-campaign art direction) | | research/icp-research.md | research-icp | Optional — audience visual preferences |

Pre-Dispatch + Mode

Canonical Pre-Dispatch: [references/_shared/pre-dispatch-protocol.md](references/_shared/pre-dispatch-protocol.md). Dimensions: asset type · downstream route · brand ref (hard-gate auto) · copy/headline · constraints. Hard-gate semantics + Cold Start + Write-back + Step 0.5 Route Detection: [references/procedures/pre-dispatch.md](references/procedures/pre-dispatch.md).

Mode ([references/_shared/mode-resolver.md](references/shared/mode-resolver.md)): budget: standard. --fast collapses Layer 1 → single pass, Layer 1.5 → 1 brief, skips Approval Gate 1 (the SELECTION pick — 3 concepts → 1; [options-selection.md](references/shared/options-selection.md), tier B). Does NOT skip hard gate, Cold Start, Step 0.5, Approval Gate 2, critic, or AI-Aesthetic Detector. Session execution profile (single-vs-multi): inherit per references/_shared/execution-policy.md.

Routes + Downstream Handoff

Every brief carries a downstream_route tag — image-gen / vector-tool / designer-handoff / template-pack. Auto-detection by asset type: [references/asset-types.md](references/asset-types.md). Override with --route=.... Per-route renderers + dispatch graph: [references/agent-manifest.md](references/agent-manifest.md).

Artifact Contract

  • Paths: docs/forsvn/artifacts/marketing/design-briefs/[slug].md (default; preserve-history → [slug].v[N].md; Gate-2 reject → [slug]-rejected.md; Gate-1 stop → [slug]-candidates.md).
  • Lifecycle: pipeline — re-run on BRAND.md / DESIGN.md update, new ASSETS.md row, brief-landing-page slot request, campaign launch, render dissatisfaction.
  • Frontmatter: skill, version, date, status, decision_state, review_tool, reviewed_at, reviewer, downstream_route, target_tool (when image-gen), asset_type, platform, dimensions, brand_anchors, sacred_respected.
  • Consumed by: image-gen tools (Claude Design / Midjourney / Imagen / DALL·E / Ideogram / Veo / Suno) · vector tools (Pencil / Figma) · human designers · brief-landing-page (LP slot) · brand/ASSETS.md auto-tick (literal path match).
  • Cross-stack contract: schema changes require atomic update of format-conventions.md § Frontmatter + § Body section order + § Downstream Handoff Block schemas.

Full template + per-route Downstream Handoff Block schemas + ASSETS.md auto-tick: [references/format-conventions.md](references/format-conventions.md).

Anti-Patterns

[references/anti-patterns.md](references/anti-patterns.md) — Process & Approval-Gate + Brand & Token + Cross-Cutting (Claude-Design-as-brand-input, hard-gate-bypass, frontmatter-drift, wrong-skill-for-intent). Re-read before any brief ships.

Durable Rules (protected)

Completion Status

  • DONE — brief approved, critic PASS, artifact written.
  • DONEWITHCONCERNS — approved with flagged issues in frontmatter (e.g. a brief drew a [verify]-flagged platform field that wasn't confirmed against a current source, or a per-vendor physical spec the operator hasn't confirmed).
  • BLOCKED — user rejected at a gate, or external dependency missing.
  • NEEDS_CONTEXTBRAND.md or DESIGN.md missing.

Execution

At brief-binding, bind the image target tool (recorded as target_tool frontmatter on image-gen routes) — inherit tool_targets or ask once per references/_shared/tool-target.md; the image-gen prompt tunes to it, tool-agnostic stays the default. Offer the registry-gated fork (category image) — Brief-only: hand the approved brief to the renderer named by downstream_route (→ produce-assetevaluate-asset); Assisted/Direct: render via a verified engine, you approve at the gate. See [execution-fork.md](references/shared/execution-fork.md); record execution_mode. On render dissatisfaction, re-invoke — pipeline lifecycle overwrites. When picking Assisted/Direct, batch-check the render surface first (all blockers at once + named fallback) — see [capability-preflight.md](references/shared/capability-preflight.md).

Worked Example

End-to-end OG-image walkthrough (image-gen route, Layer 1 → 3 concepts, Gate 1 pick, Layer 2 prompt-craft, critic PASS, Gate 2, ASSETS.md auto-tick) + cycle-2 FAIL + --fast variant + Route C snippet: [references/examples/brief-graphic-walkthrough.md](references/examples/brief-graphic-walkthrough.md). Additional asset-type examples: [references/examples.md](references/examples.md).

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.