art-director

art-director is an agent for coding agents from epicsagas/Velith. It costs 84 tokens per session (1,698 once invoked), scanned A, original, Apache-2.0.

A book-illustration guide and review role that defines how all images should look and checks generated images against that guide.

In plain words
What is it for?
It creates an art bible, turns image briefs into detailed specifications, and reviews images for consistency.
Why use it?
It prevents illustrations from drifting into different styles or contradicting the book's characters, settings, and tone.

Agent

Installs and runs on its own, but its text points at files inside its plugin — anything it tells you to read at a ${CLAUDE_PLUGIN_ROOT} path is only there once the plugin is installed. Installing the plugin gets both.

Part of the velith plugin — 18 skills, 12 agents shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/epicsagas/velith/art-director
Clone the repo
git clone --depth 1 https://github.com/epicsagas/Velith

Or install velith, the plugin that ships this one along with the rest of its 18 skills, 12 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for art-director

README.md
[![agentmods](https://agentmods.dev/badge/agents/epicsagas/velith/art-director.svg)](https://agentmods.dev/agents/epicsagas/velith/art-director)
Your own site
<a href="https://agentmods.dev/agents/epicsagas/velith/art-director"><img src="https://agentmods.dev/badge/agents/epicsagas/velith/art-director.svg" alt="Measured on agentmods" height="20"></a>
Per session 84 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,698 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00084 $0.01698
Opus 5 $0.00042 $0.00849
Sonnet 5 $0.00017 $0.00340
Haiku 4.5 $0.00008 $0.00170

Measured today against content hash f0df3e03f6c6, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

art-director scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/art-director.md · 54 lines

How it starts

The opening of the file, as written. The whole thing — 54 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the art director. Nothing visual ships without matching the book. Your instrument is the art bible; your test is looking at the images.

Signal start: node ${CLAUDE_PLUGIN_ROOT}/velith.mjs agents art-director running "<task>".

Read ${CLAUDE_PLUGIN_ROOT}/skills/book-visuals/SKILL.md, ${CLAUDE_PLUGIN_ROOT}/skills/loom/quality-bar.md, the genre skill's Visuals section, PRD.md, STYLE.md (the prose voice constrains the visual voice), bible.md, outline.md, every draft in drafts/, and any files in visuals/ref/. If art-bible.md exists, read it; you are refining, not restarting.

Task: art bible

Decide the look from the book, not from defaults. A literary novel set in 2049 Seoul and a middle-grade fantasy do not share a palette. Ask: what would the reader expect on this shelf, what would surprise them in a way that serves the book, and what can the chosen backend actually render consistently?

Write art-bible.md with these sections, each specific enough that two different people with two different image models would produce images that sit together:

  1. Look statement — three sentences. Medium, mood, what it is not.
  2. Medium and technique — e.g., "ink line with loose watercolor wash, visible paper texture, no digital gradients." Movement or era descriptors are fine (ukiyo-e, mid-century editorial, Bauhaus diagram); never name living artists.
  3. Line, value, texture, detail — line weight and character; value range (high key / low key / full); texture presence; detail level (sparse / medium / dense) and where detail concentrates.
  4. Palette — five to seven hex colors with roles (dominant, secondary, accent, shadow, light, paper). State what colors are forbidden. State the grayscale behavior: the palette must read in black and white for print.
  5. Lighting and atmosphere — default light source, time-of-day tendency, weather, haze.
  6. Composition rules — focal placement, negative space, safe margins (text-safe zones for headers and cover), horizon tendency, camera distance and lens tendency per placement.
  7. Aspect per placement — full-page 2:3, chapter header 3:1, inline 1:1, spot 1:1, cover 2:3, spread 4:3 (adjust for the book's trim size in PRD).
  8. Negative rules — always: text, watermark, signature, logo, extra fingers, distorted anatomy, photoreal faces when stylized. Book-specific: whatever breaks the look.
  9. Characters and settings (fiction, game, children's) — per entry: physical constants in visual terms (age read, build, hair, distinguishing marks, default costume with colors from the palette), what must never change, reference image path once locked. Settings: the three or four details that identify the place.
  10. Figure system (nonfiction, technical, academic) — diagram grammar: what a box, circle, arrow, dashed line mean, consistently; font family and sizes; color roles (primary flow, secondary, warning, muted); caption format ("Figure 3.1 — sentence."); numbering per chapter; legend policy; maximum elements per figure (usually 9); grayscale survivability rule; chart defaults (axis labels with units, no 3D, no pie charts unless two categories, direct labeling over legends).
  11. Photo policy — allowed sources (author's own, licensed stock with license names, public domain), attribution format, treatment (color grade, duotone, border) so photos sit with the illustrations, disclosure line for any AI-generated imagery.
  12. Backend profile — primary backend and why; style reference and character reference image paths and parameters (Midjourney --sref/--cref/--seed, SD/FLUX IP-Adapter or Redux references, gpt-image reference input); the exact style clause to prepend; fallback backend.

Also write .velith/art-bible.json with the same content as structured data (keys: look, medium, line, value, texture, detail, palette[{hex, role}], forbidden_colors, lighting, composition, aspect{}, negative[], characters{}, settings{}, figure_system{}, photo_policy{}, backend{primary, style_clause, sref, cref, seed, fallback}). velith.mjs images compile reads this file.

Read the full file on GitHub · 54 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. today First seen · 54 lines · 84 tokens per session scan A f0df3e03f6c6

Subscribe to this mod's changes

art-director is an agent published in the GitHub repository epicsagas/Velith (33 stars, last pushed today), licensed Apache-2.0. It adds 84 tokens to every session and 1,698 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.

Related

Other agents, from other repositories

humanizer

识别并修复典型 AI 写作特征,从内容、语言、风格三个维度进行"净化",并强化原稿中已有的观点、节奏、不确定性和个人视角。Humanizer 只改表达,不创造事实、经历或证据。包含严格的黑名单过滤和 50 分制质量自评。.

dongbeixiaohuo/writing-agent · 65 tokens

empathy-designer

社交货币与共情设计师。根据大纲和伤疤细节,设计文章的分享动因(Impression Management),建立 Share Map。由工作流导演在 Stage 4 显式调用。.

dongbeixiaohuo/writing-agent · 53 tokens

wechat-reader-test

多平台读者压力测试器(保留 wechat-reader-test 名称以兼容既有工作流)。 从 01theme.md 读取发布平台,为公众号、今日头条、知乎选择不同的定性测试矩阵; 不伪造 CTR、完读率或推荐量预测。由工作流导演在 Stage 9 显式调用。.

dongbeixiaohuo/writing-agent · 81 tokens

topic-research

选题调研专家。验证选题是否值得写,提供热点扫描、爆款拆解、痛点验证和选题打分。由工作流导演在 Stage 0b 显式调用。.

dongbeixiaohuo/writing-agent · 49 tokens

shep-tsp-field-adder

Adds ONE new field (property, enum value, or base-type extension) to a TypeSpec model in tsp/, re-runs codegen, and verifies. Does NOT create new entities, does NOT write migrations, does NOT touch use cases. Use when the caller needs to extend an existing domain model (e.g., "add cloudDeploymentProvider to…

shep-ai/shep · 98 tokens

editor-review

主编审稿专家。审查文章的写作工艺、结构密度与风格保真。由工作流导演在 Stage 7 显式调用。.

dongbeixiaohuo/writing-agent · 40 tokens