greeun

Branding Harness — Design skill for Claude Code

Design community

Claude Code skill: generate an integrated brand identity system (strategy + verbal + visual direction) with a Planner → Generator → Evaluator harness from Anthropic's Harness Design article.

How to install Branding Harness

This entry records only its repository, not the path inside it, so there is no exact command to give. Open greeun/branding-harness and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Branding Harness does

Claude Code skill: generate an integrated brand identity system (strategy + verbal + visual direction) with a Planner → Generator → Evaluator harness from Anthropic's Harness Design article.

Alternatives in Design

Brand Guidelines — Apply Anthropic's official brand colors and typography to artifacts 94.1k ★
CloudCLI — 8,400+ Free open-source web/mobile UI for Claude Code, Cursor CLI, and Codex 8.7k ★
Pair — Fuse two extracted designs into a single hybrid identity 3.6k ★

README

Branding Harness — v0.1.0

[한국어 README](./README.ko.md)

A Claude Code skill that turns a rough company/product/service description — or a raw business idea — into a complete, internally-consistent **integrated brand identity system**, using the **Planner → Generator → Evaluator** harness pattern from Anthropic's [Harness Design for Long-Running Application Development](https://www.anthropic.com/engineering/harness-design-long-running-apps) (Prithvi Rajasekaran, 2026).

What It Does

Given a brand brief, the skill produces one assembled brand guideline document spanning four layers:

Layer	Deliverables
Strategy	Brand core (mission/vision/values), target-audience definition, competitive differentiation, positioning statement (template-locked), brand archetype (12-Jung framework), value proposition
Verbal	Naming candidates + screening notes, tagline, voice & tone guide (≥3 principles with do/don't pairs), core message framework (elevator pitch / one-liner / boilerplate)
Visual direction	Color, typography, logo, imagery text directives — not rendered images — each back-traceable to a strategy attribute
Assembly	The brand guideline document + a decision↔strategy-pillar traceability table with zero orphan decisions

⚠️ This skill outputs **text directives**, not rendered logo/visual images. Image-generation requests are out of scope and get redirected.

Why a Harness?

The target model (Sonnet 4.5) defaults to generic, undifferentiated "AI slop" branding and shows strong context anxiety (wraps up early). A harness structurally prevents two failure modes:

Self-evaluation bias — when grading its own work, an LLM "confidently praises mediocre output." Generation and evaluation run as separate subagents that never see each other's reasoning.
Context anxiety — as context fills, the model rushes to finish. Fixed small sprints + decisive context resets (fresh sessions