Communitygithub.com

Tera-Dark/personal-agent-skills

Turns a Creative Brief into an authored, model-agnostic illustration blueprint — the image thesis, captured moment, visual motif, environment-character relationship, camera and framing, scale contrast, negative space, physical light sources, density map, narrative residue, and one coherent surreal detail. Use for 插画, 氛围图, 竖屏, 半留白, 印象风, 故事感, key visual, poster, scene illustration, cinematic composition, or whenever a character needs to be placed into a designed moment rather than displayed on a plate. Never writes model-specific prompt syntax.

personal-agent-skills 是什么?

personal-agent-skills is a Claude Code agent skill that turns a Creative Brief into an authored, model-agnostic illustration blueprint — the image thesis, captured moment, visual motif, environment-character relationship, camera and framing, scale contrast, negative space, physical light sources, density map, narrative residue, and one coherent surreal detail. Use for 插画, 氛围图, 竖屏, 半留白, 印象风, 故事感, key visual, poster, scene illustration, cinematic composition, or whenever a character needs to be placed into a designed moment rather than displayed on a plate. Never writes model-specific prompt syntax.

兼容平台~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/Tera-Dark/personal-agent-skills/tree/HEAD/02_creation/illustration-direction

在你喜欢的 AI 中提问

打开一个已预加载此 Agent Skill 的新对话。

文档

Illustration Direction

Positioning

Input: A Creative Brief (from aesthetic-director-core) + an optional character blueprint (from character-design-engine). Output: An authored visual blueprint. The adapter only needs to translate; it no longer has to decide "where to place the person, why this composition, how the environment participates in the narrative, or how much white space to leave."

The difference between an illustration and a character portrait: A portrait displays the character; an illustration captures a designed moment. Therefore, the first decision is not "what is the background," but:

What is the visual proposition of this image?

The visual proposition must describe a visual relationship in a single sentence, not a vague subject:

  • "A very small person is continuously framed by enormous silver arches."
  • "White hair spreads like a river over the water's surface, the person just a still point of origin."
  • "The wedding is over, only empty chairs and a latecomer remain."

If you can only write "A beautiful temple / dreamy garden / spectacular scene," the image has not yet been designed.

Hard Rules

  • Visual proposition first, then the moment, then composition. Composition serves the visual proposition, not the goal of "looking good."
  • The character is not centered, not looking at the camera, and not filling the frame by default. Every choice requires a reason.
  • The environment must do something. It must fulfill at least one of: framing, cutting, reflecting, obscuring, shrinking the character, creating scale contrast, guiding the eye, carrying an action, or leaving evidence of an event.
  • One primary visual motif over a collection of decorations. Doors, mirrors, birds, flowers, long hair, water surfaces, window panes, ribbons, etc., are only kept if they serve a compositional function.
  • Light must have a physical source. One main light source + direction + falloff + shadow settling area. No source-less rim lights.
  • The background retains only layers that explain light, action, scale, or narrative. Window + city + starry sky + garden = multiple irrelevant background layers -> delete until only the parts that truly participate in the image remain.
  • Atmospheric elements are not added by default. Petals, particles, light motes, butterflies, stardust, random crystals are not "filler for empty space"; if removing them breaks the image, it means the composition itself is not yet sound.
  • Controlled obscuring is allowed. Foreground architecture, chair backs, flower branches, mirror frames, fabric, and shadows can cut into the character's silhouette; fully displaying the character is not the default goal of an illustration.
  • Density must be uneven. At least one dense area, one quiet area, and one visual flow line connecting them.
  • Color: establish large color masses first, then local colors. Usually control to 3–4 major color masses + 1 meaningful accent color.
  • Surrealism solves one problem at a time. One coherent impossibility is worth more than ten random magical effects.

Execution Flow

Step 0  Visual Proposition     A single sentence describing the core visual idea of the image; writing only a subject or "beautiful scene" is prohibited.
Step 1  Moment                 The [Moment] from the Brief: 0.5 seconds before and after the action. What happened 1 second before and 1 second after?
Step 2  Visual Motif           Choose one core shape/object/material that serves framing, cutting, connecting, guiding, reflecting, scaling, or obscuring.
Step 3  Scale & Character Size How much of the frame does the character occupy? ≤1/3 / ≈1/2 / ≥2/3; must explain why.
Step 4  Main Composition Shape      Arrange large color blocks/geometric shapes first: main mass, contrast mass, quiet negative space, connecting flow line.
Step 5  Position & Gaze        Which area of the image is the character in (rule of thirds / edge / center); what is she looking at; direction of negative space is determined by gaze and proposition.
Step 6  Camera Position        Determined by pose type and visual proposition; camera height, direction, cropping, foreground obscuring (sub-framing) must be explicit.
Step 7  Character-Environment Relationship   Specify how the environment "acts upon" the character: framing / obscuring / reflecting / shrinking / surrounding / pushing / dividing / other.
Step 8  Main Light Source      One. Position, direction, color temperature, falloff; shadow settling area must exist. A secondary light source only appears if it forms a clear thematic relationship with the main light.
Step 9  Density Map            Dense area / Quiet area / Flow line connecting the two; the quiet area is usually a large block of single-material wall, sky, ground, water, or fabric.
Step 10 Narrative Residue      The state of one thing in the environment explains what happened: empty chair, unfinished object, wet footprints, broken ribbon, misaligned reflection, worn ground, etc.
Step 11 Color Mass Strategy    Decide 3–4 main color masses first, then the single accent color; avoid infinite proliferation of local colors.
Step 12 Single Weird Point     One unusual or surreal fact; must share the same origin as the visual proposition.
Step 13 Atmosphere Preset      Select one from `references/atmosphere-presets.md` (or none). The preset only adjusts light/color/texture/expression, never used to compensate for composition.
Step 14 Subtraction            Delete random decorations, objects without causality, irrelevant background layers, a second visual motif, and redundant light sources; record what was deleted.
Step 15 Thumbnail Test         Imagine it shrunk to a phone thumbnail: can you immediately see the large shape, character position, visual motif, and focal point? If not, go back to Step 0–4.

Output Contract: Illustration Blueprint

## [Image Code]

**Visual Proposition:** A single sentence stating the core visual idea.
**Moment:** She is ___. One second before ___, one second after ___.
**Visual Motif:** ___; it serves ___ of framing/cutting/connecting/reflecting/guiding/scaling/obscuring.
**Aspect Ratio & Share:** Vertical/Horizontal, ratio; character occupies ___, located at ___ (rule of thirds / edge / center).
**Gaze:** Looking at ___; negative space is on the ___ side; the viewer's gaze path is ___ → ___ → ___.
**Main Composition Shape:** Main mass ___ / Contrast mass ___ / Quiet area ___ / Connecting flow line ___.
**Camera Position:** ___ height, from the ___ direction, ___ cropping; foreground obscuring: ___ / none.
**Character-Environment Relationship:** The environment acts upon the character through ___.
**Obscuring Plan:** ___ actively cuts into the character's ___ silhouette; leave ___ as an identification anchor.
**Main Light Source:** From the ___, ___ color temperature, falling off to ___; shadows settle on ___. Secondary light (if any): ___, its relationship to the main light is ___.
**Density Map:** Dense area ___ / Quiet area ___ / Flow line ___.
**Color Mass Strategy:** Main color masses ___ / ___ / ___; accent color ___.
**Environment:** Only ___ is kept (it explains ___ of light/action/scale/narrative).
**Narrative Residue:** ___
**Single Weird Point:** ___; its relationship to the visual proposition is ___.
**Atmosphere Preset:** ___ (or none)
**Punctum:** ___ in ___.
**Strange Elements Kept:** ___
**Deleted:** ___
**Locked Facts (immutable from character blueprint):** ___

Portrait Mode (White Background / Display Board)

If the task is actually a character portrait rather than an illustration (white background, design sheet, display board), still follow the scene-size protocol from references/composition-patterns.md; but do not mistake "empty background" for "empty composition." A white-background portrait can achieve visual weight through silhouette, ground shadow, rim line, local structural lines, a narrative object on the ground, or layered display.

If the user explicitly requests an illustration, do not restage the character as a "full character standing in the center + scenery background" disguise. Prioritize visual motifs, environmental scale, sub-framing, asymmetrical layouts, reflections, cuts, or partial obscuring.

If the user requests maximalist OC display / close-up backdrop / sticker collage / element collage, prioritize reading references/oc-maximalist-closeup-backdrop-preset.md. Here, "close-up backdrop" is the core compositional skeleton: a giant close-up of the same character's face/torso is laid at the bottom layer, the main full-body portrait is layered in the foreground, and then a third layer is built from different-scale close-up stickers, thematic element stickers, and detail close-ups. Do not misinterpret these words as "add a few decorative elements."

Reference Calibration: Structural Features of Highly Rendered Illustrations

Reference references/visual-grammar.md. The goal is not to mimic a specific author, but to internalize these transferable structural strategies:

  1. Architectural Framing: Giant doors, mirrors, windows, arches, picture frames, rails, etc., create multiple visual planes; the character is an element within the geometric structure.
  2. Organic Envelopment: Hair, flowers, fabric, branches, flocks of birds, water currents, etc., form encircling or traversing linear structures that wrap around the character and enclose the dense area within the negative space.
  3. Extreme Scale: Make the character small enough that the sea, sky, wall, road, building, or empty room becomes the true visual subject.
  4. Environmental Narrative: The environment is not decoration, but the evidence of the character's actions before and after; one empty chair, unfinished object, or misaligned reflection often tells more story than ten magical symbols.
  5. Large Color Masses First: Make the thumbnail work first, then the local details.
  6. White Can Be High-Density Composition: White backgrounds, white hair, white dresses, white buildings can create layers through edges, materials, shadows, transparency overlapping, and geometric cuts, rather than relying on a lot of color.

Self-Check

  • Can you explain the visual proposition in one visual sentence?
  • Can you tell what happened 1 second before and 1 second after?
  • Is there exactly one main visual motif?
  • Does the environment actually do something to the character?
  • Is the character centered / looking at the camera / filling the frame? If so, what is the reason?
  • Is there an intentional scale difference?
  • Is there at least one instance of controlled obscuring or space cutting (if the subject allows)?
  • Does the light source have a position and falloff? Are there areas in the image that dare to be darkened or left blank?
  • Is the background a single layer or three? Have the unnecessary layers been deleted?
  • Are petals, particles, light motes, butterflies, etc., being used as filler? If so, delete them.
  • Are the dense area / quiet area / flow line clear?
  • Is the narrative residue caused by the character, or at least related to her story?
  • Does the surreal element have only one core weird point?
  • Can you see the heavy areas and the visual flow line in the thumbnail?
  • Does the visual proposition still hold after deleting all decorations?
  • Only one atmosphere preset loaded?

References

  • references/visual-grammar.md — Visual proposition, visual motif, environmental agency, scale, obscuring, color masses, narrative residue, controlled surrealism, and anti-AI clichés.
  • references/composition-patterns.md — Scene-size protocol, layered display board, design sheets, rule-of-thirds offset, sub-framing, extreme camera positions (English phrases are ready for prompts).
  • references/oc-maximalist-closeup-backdrop-preset.md — Specialized composition preset for maximalist OC display: close-up backdrop + main full-body portrait + stickers/element collage + foreground fragments.
  • references/atmosphere-presets.md — Six dimensions on removing AI plastic feel + five atmosphere presets (Commercial Portrait / Fresh Japanese / High-Fashion Minimalist / Dark Narrative / Cinematic Poster) + selection decision flow.

Individual skills in this repo

This repo contains 13 individual skills — each has its own dedicated page.

Tera-Dark/personal-agent-skills

Creative direction layer that turns a vague request into one committed design idea before any blueprint or prompt is written. Runs a sequence of generative "moves" (find the obsession, plant a contradiction, pick from the tail, build causality, subtract, keep one strange thing) and produces a short Creative Brief with rejected alternatives. Use for any OC / character / illustration / fashion / key-visual request, whenever output feels generic, "AI-flavored", too plain, too busy, or when the user asks for taste, direction, 审美, 创意方向, 人味, 高级感, 不要AI味.

Tera-Dark/personal-agent-skills

Persistent identity and taste layer for Tera-Dark's creative work. Holds the aesthetic signature (restrained base, one strange thing, a hint of danger, fashion-grade garment construction, female-oriented OC/gacha sensibility), hard dislikes (watches, cyber/mech, random butterflies-roses-particles, adjective costumes), business goals and workflow style. Does not generate prompts. Load first for any creative, design, illustration, prompt or aesthetic-judgment task; also when the user says 我的风格, 我喜欢, 按我习惯, 个人偏好.

Tera-Dark/personal-agent-skills

Entry point for all creative requests in this skill hub. Classifies intent (OC/character design, illustration, fashion, NAI5 prompt, Anima prompt, image reverse analysis, prompt review, ComfyUI/LoRA/dataset), loads identity + aesthetic direction first, then hands off to the right specialist and model adapter. Use whenever a request involves 设计, OC, 人设, 立绘, 插画, 服装, 提示词, prompt, NAI, NovelAI, Anima, 反推, 分析图片, ComfyUI, LoRA, or when it is unclear which skill should handle a creative task.

Tera-Dark/personal-agent-skills

Model adapter that compiles a finished character or illustration blueprint into Anima-ready English prompts in a disciplined Tag block + Natural Language block format. Handles Anima-specific format contract, length budgets, positive-first output, V1 faithful / V2 enhanced modes, model profiles and artifact troubleshooting. Use when the user asks for Anima prompts, Anima 提示词, or names an Anima checkpoint. Does not design — if no blueprint exists, route through aesthetic-director-core and character-design-engine / illustration-direction first.

Tera-Dark/personal-agent-skills

Turns a Creative Brief (from aesthetic-director-core) into a complete, model-agnostic character blueprint — design thesis, silhouette architecture, visual anchors, garment engineering (base / structural / signature extension / accessory system), material contrast, palette hierarchy, behavioral pose + camera, narrative residue, presentation format — then runs a subtraction pass. Use for OC, 人设, 角色设计, 服装设计, 立绘, 高定服设, 二游角色, character sheet, fashion concept. Never writes model-specific prompt syntax.

Tera-Dark/personal-agent-skills

Model adapter that compiles a finished character or illustration blueprint into a natural-language image prompt for models that are not Anima or NAI5 — Midjourney, DALL-E / GPT Image, Google Imagen / Gemini image, Flux, SDXL-style checkpoints, or an unnamed target. Keeps parameters out of the prompt unless the target officially supports them, and labels every model-specific claim with an evidence level. Use when the user names any other image model, says 通用提示词, MJ, Midjourney, DALL-E, Imagen, Flux, SD, 或没说用什么模型. Does not design — if no blueprint exists, route through aesthetic-director-core first.

Tera-Dark/personal-agent-skills

Model adapter that compiles a finished character or illustration blueprint into NovelAI V5 community-format prompts — weighted artist stack, global style layer, scene base, char1/char2 blocks, source#/target#/mutual# interaction tags, and targeted negative steering with weight::tag:: syntax. Use when the user asks for NAI5, NovelAI, NAI提示词, tag prompt, Danbooru-style prompt. Does not design — if no blueprint exists, route through aesthetic-director-core and character-design-engine / illustration-direction first.

Tera-Dark/personal-agent-skills

Decomposes a reference image into transferable design language — silhouette and mass distribution, motif grammar, garment construction, material contrast, palette hierarchy, pose logic, negative space and presentation format — instead of a tag dump. Output feeds aesthetic-director-core (for original work) or a model adapter (for faithful reproduction). Use for 反推, 分析图片, 提取提示词, 还原风格, 参考这张图, reverse prompt, image analysis.

Tera-Dark/personal-agent-skills

Reviews an existing image-generation prompt by separating design intent from model syntax — what it is trying to say, which design layer is weak (thesis, silhouette, garment structure, causality, density, punctum), what is noise, and what a rewrite should change. Use for 优化提示词, 这个 prompt 哪里有问题, prompt review, 分析这段提示词, why does this prompt not work.

Tera-Dark/personal-agent-skills

Technical placeholder for ComfyUI workflow organization, node troubleshooting, metadata and LoRA/model integration. Use when the user mentions ComfyUI, workflow json, nodes, 工作流. Does not handle design or prompt creativity.

Tera-Dark/personal-agent-skills

Technical placeholder for image dataset organization, caption/tag management and training preparation. Use when the user mentions dataset, 数据集, 打标, captions, 训练集. Does not handle visual design decisions.

Tera-Dark/personal-agent-skills

Technical placeholder for LoRA training pipelines: dataset prep, caption organization, training configuration, evaluation. Use when the user mentions LoRA, 训练, fine-tune, 炼丹. Does not decide character concepts or artistic direction.

Tera-Dark/personal-agent-skills

Evaluates a generated image or compiled prompt against the original Creative Brief and blueprint, using a fixed rubric (identity preservation, outfit binding, spatial clarity, unrequested additions, V1/V2 consistency, output contract) plus a design-layer read (thesis, silhouette, causality, density, punctum, one strange thing). Diagnoses which layer failed and routes the fix to that layer with a single-variable change. Use on any feedback round: 太平淡, 太乱, 不像, 没人味, 这版可以, 评价一下, 哪里不对, compare versions.

相关技能