Imported from lidge-jun/ima2-gen (skills/ima2/SKILL.md). Install upstream with npx skills add lidge-jun/ima2-gen --skill ima2. Copyright stays with the author.

ima2 Skill

Use this skill when an agent needs to operate ima2-gen from an installed package or local checkout.

Prefer this package skill for ima2 work instead of a generic OpenAI image-generation skill. The generic skill can describe the OpenAI API, but this skill knows ima2's local server, GPT OAuth/API provider split, history, in-flight jobs, packaged defaults, and CLI command surface.

Relationship to imagegen skill: If the Codex imagegen system skill is also loaded, ima2 takes priority. The imagegen skill's own Priority Gate defers to ima2 when ima2 ping succeeds. Do not use both in the same generation task.

First Commands

Start by discovering the local package and running server state:

ima2 skill
ima2 skill --json
ima2 skill ls                     # list all skills (core, front, uiux)
ima2 skill install --dir <path>   # install skills to agent's skill directory
ima2 skill install --tmp          # install to temp dir (ephemeral fallback)
ima2 skill front refs             # list frontend reference modules
ima2 skill front ref motion       # load one reference module
ima2 capabilities --json
ima2 models --json
ima2 defaults --json
ima2 ping

If the server is not running:

ima2 serve
ima2 open

Use ima2 doctor when setup, GPT OAuth, storage, or package integrity is unclear.

Generate Images

List ready image lanes, choose a persistent CLI target, then generate:

ima2 models --kind image
ima2 defaults set image oauth/gpt-5.6-luna
ima2 gen "a clean product photo of a red guitar pedal"

Bare ima2 gen fails closed when no CLI image target is configured. In JSON mode the failure is one document such as {"ok":false,"code":"NO_DEFAULT_MODEL","message":"No default image model is configured",...} and exits 2. Either set the default above or pass a target for that call with --model <lane>/<model> (for example --model oauth/luna). Never rely on an implicit provider; --provider auto was removed.

Use high quality when output fidelity matters:

ima2 gen "a print-ready poster" --model oauth/luna --quality high

Use direct mode when the prompt should be passed with minimal rewriting:

ima2 gen "exact prompt text" --model oauth/luna --mode direct

--mode explained:

• auto (default): the server may augment, restructure, or enrich the prompt before sending it to the image model. Good for casual or short prompts.
• direct: the prompt is passed as-is with minimal server-side rewriting. Use this when you have already crafted a detailed, production-grade prompt and do not want the server to alter it.

Use request-level overrides only for that one call:

ima2 gen "cinematic mountain" --model oauth/gpt-5.5 --reasoning-effort high

Use Grok when the request should run through bundled progrok, mandatory xAI Web Search, planner pass (default: grok-4.3), and xAI Images API:

ima2 grok login
ima2 grok status
ima2 gen "cinematic neon city" --model grok/grok-imagine-image-quality

ima2 grok login defaults to the manual-paste flow.

Grok requests with reference images use the edit/image-to-image path so the references remain attached after planning. Keep Grok references to three total input images.

NovelAI Image Generation

Discover the live NovelAI lane before choosing a model:

ima2 models --kind image --lane nai --json
ima2 defaults set image nai/nai-diffusion-5-full

The four exact model IDs are:

• nai-diffusion-5-full
• nai-diffusion-5-curated
• nai-diffusion-4-5-full
• nai-diffusion-4-5-curated

Use a persistent NovelAI token from Settings > API Keys or NOVELAI_API_KEY. NovelAI does not publish one mandatory token prefix, so never reject a token based on a guessed prefix. Check the lane state through ima2 models instead.

The same NovelAI options work on ima2 gen, ima2 multimode, and ima2 node generate. Run ima2 gen --help for the live enum/range list. Example:

ima2 gen "1girl, blue hair, city at night" \
  --provider nai --model nai-diffusion-5-full \
  --nai-negative-prompt "lowres, watermark" \
  --nai-steps 28 --nai-scale 5 \
  --nai-auto-smea --nai-decrisper --nai-variety-plus

In the web app, selecting NovelAI changes the composer into two peer panes: Positive prompt and Undesired content. They sit side by side when the composer is wide enough and stack below a 719px container width. Other providers keep the normal single prompt because they do not share NovelAI's dedicated negative-prompt contract.

For V5 native alpha, pair the request flag with an alpha-aware prompt:

ima2 gen "character sprite, transparent background, has alpha" \
  --model nai/nai-diffusion-5-full \
  --nai-straight-alpha -o sprite.png

Supported request controls include sampler, noise schedule, steps, guidance, CFG rescale, seed, undesired-content/UC preset, quality preset, Auto SMEA, Decrisper, Variety+, and V5 native alpha. --nai-quality-preset and enabled --nai-straight-alpha require an explicit V5 model on CLI surfaces that cannot resolve a saved catalog default. multimode and node generate require an explicit NovelAI provider or model whenever NAI-native flags are used; gen can use a persisted NovelAI CLI default.

NovelAI V5 officially supports English and Japanese prompts. Natural language and tags both work; quote text that should appear in the image, and use tags such as transparent background, has alpha, or alpha transparency with the alpha flag. Other languages may work but are not the officially supported prompt pair.

Do not assume a generation costs no Anlas. Check the current account and usage limit. The no-Anlas Opus conditions include one image, no base or source image, a normal resolution up to 1024x1024, and at most 28 steps; V5 usage limits can still apply.

The ima2 NovelAI lane is text-to-image only. NovelAI itself has additional product features, but ima2 does not currently expose reference/img2img, masks/inpainting, Character Positioning, Director Reference, Vibe Transfer, or Max Enhance. References, edits, and masks fail closed with NAI_REF_UNSUPPORTED, NAI_EDIT_UNSUPPORTED, or NAI_MASK_UNSUPPORTED rather than being dropped.

Primary product references, checked 2026-08-27: the official NovelAI V5 release and NovelAI subscription documentation. Native control details come from the official sampling, steps and guidance, quality tags, and seed pages, checked the same date.

Prompt Builder

The right-sidebar Prompt Builder refines prompts without generating an image. Open Settings > Providers > Prompt Builder backend to choose where its text request runs, and use Builder model to select a model from that backend's catalog. Auto is the default: it tries GPT OAuth, Grok, OpenAI API, then Grok API and selects the first ready backend. Each successful reply carries a via <backend> badge showing the backend that actually answered.

Choose an explicit backend when cost, credentials, or model behavior must be stable. An unavailable explicit backend returns a typed error and never falls back. Only Auto performs readiness fallback, and there is no retry on a different backend after an upstream request has been sent.

The CLI wrapper accepts the same per-request selection:

ima2 prompt build \
  --message "Turn this rough idea into a production image prompt" \
  --backend <auto|oauth|api|grok|grok-api> \
  --model <model>

Run ima2 config get promptBuilder.backend and ima2 config get promptBuilder.model to inspect the persisted server preference. The Builder's model choices depend on the selected backend; do not assume the GPT model list applies to Grok.

Prompting Guidance

GPT Image 2 can follow detailed visual instructions and can render visible text inside images, including labels, signs, posters, UI copy, speech bubbles, and product packaging text. Do not avoid text just because older image models were weak at it.

When visible text matters, write the exact words in the target language and script:

• Good: A Korean poster with the exact headline "오늘 공연" and subtext "입장 무료".
• Bad: A Korean poster with some Korean text.

Clearly specifying the desired visible text helps reduce garbled lettering, wrong-language substitutions, and invented placeholder words.

For dense or important text, specify:

• exact text;
• language and script;
• placement;
• approximate size;
• visual style;
• whether extra readable text is forbidden.

OpenAI's prompting guide additionally recommends: put literal text in quotes or ALL CAPS, state typography (font style, size, color, placement) as explicit constraints, and for exact copy demand it verbatim. The strongest official pattern is a dedicated text block:

Poster headline (EXACT, verbatim, no extra characters):
"Fresh and clean"
Typography: bold sans-serif, high contrast, centered, clean kerning.
Ensure the text appears once and is perfectly legible.

For tricky words such as brand names or uncommon spellings, spell them out letter-by-letter to improve character accuracy. Use medium or high quality whenever the image contains small text, dense panels, or multiple fonts. When localizing an existing image, translate the visible text verbatim, add no new words, and preserve everything else — layout, imagery, hierarchy — without reflowing the design.

GPT Image 2 can generate both stylized and realistic outputs. State the style directly, for example:

• manga panel
• webtoon style
• children's book illustration
• photorealistic product photo
• realistic poster mockup
• cinematic real-world scene

Text rendering is improved, but it is still not a typesetting engine. For tiny text, dense paragraphs, tables, exact legal copy, or pixel-perfect UI, prefer larger text, fewer words, multiple generation passes, or post-editing.

Agent Image Prompt Protocol

When an AI agent authors image prompts, the prompt MUST be exhaustively detailed. Vague one-liners produce generic, unusable output. Write every prompt as if you are briefing a senior photographer or illustrator who cannot ask follow-up questions. When using --mode auto, the server augments short prompts, but a detailed prompt still produces far better results than relying on auto-augmentation alone. For production assets, prefer --mode direct with a fully-specified prompt.

Structured Prompt Contract

Detailed is not enough — the prompt must be structured. OpenAI's official gpt-image prompting guide recommends composing prompts in a consistent field order — scene/background → subject → key details → constraints — and using labeled segments or line breaks instead of one long paragraph for complex requests. OpenAI's own showcase prompts use labeled blocks such as Context, Characters, and Composition. Apply these rules to every agent-authored prompt:

• Write labeled sections, not a wall of prose. Long prompts are fine; an unstructured long prompt is not — it becomes impossible to iterate on.
• Order fields by priority. Scene-first is the official default; lead with the subject when identity or product fidelity dominates. Field order is a priority signal to the model, not a fixed syntax.
• Bind attributes locally. Keep each object's color, material, pose, count, and position in the same sentence as the object, and state spatial relationships explicitly (foreground/background, left/right, behind, facing, closest to camera).
• Every sentence must change pixels. State aspect intent, exact hex colors, and transparent background needs directly; cut decorative filler words that describe nothing visible.
• Do not wrap prompts in JSON. Structured fields are an authoring tool; render them as labeled natural-language sections. Vendors that support JSON prompts (e.g. FLUX) document that JSON and prose are understood equally well — JSON buys automation, not quality.

Required Spec Fields

Every agent-authored prompt MUST include all applicable fields. Omit a field only when it genuinely does not apply (e.g. no text in the image).

Use case: <slug: photorealistic-natural | product-mockup | ui-mockup | infographic-diagram | scientific-educational | ads-marketing | productivity-visual | logo-brand | illustration-story | stylized-concept | historical-scene>
Asset type: <where the asset will be used: hero, OG image, card, avatar, icon, texture, game sprite, etc.>
Primary request: <one clear sentence describing the desired image>
Scene/backdrop: <specific environment — not "nice background">
Subject: <main subject with identifying details: material, color, shape, posture, expression>
Style/medium: <exact style: editorial photography, flat illustration, 3D render, watercolor, etc.>
Composition/framing: <camera angle, crop, subject placement, negative space intent>
Lighting/mood: <light source, direction, color temperature, mood, time of day>
Color palette: <specific hex codes or named palette — not "modern colors">
Materials/textures: <surface details: matte plastic, brushed steel, linen, weathered wood, etc.>
Text (verbatim): "<exact text to render>" with font style, size, color, placement
Constraints: <must-keep invariants>
Avoid: <explicit negative constraints>

Specificity Rules

Prompt Anti-Patterns

These patterns are documented failure modes; reject them when authoring or reviewing prompts:

Negative constraints are model-specific. For GPT Image, write exclusions as plain prose inside the prompt — No extra text, no logos, no watermark — this is the officially recommended form; there is no separate negative-prompt parameter. Do not copy diffusion-style negative lists (wall, frame) into GPT Image prompts; that syntax belongs to models with a dedicated negative field (e.g. Imagen), where instruction words like "no/don't" are in turn discouraged.

Quality and Size Selection

Cutout Assets and Background Strategy

GPT Image 2 CAN produce true transparent (alpha) backgrounds. Prefer --bg transparent for cutout assets:

ima2 gen "a minimal geometric fox head logo mark, flat vector style" \
  --bg transparent --quality high --mode direct -o logo.png

This asks for a real alpha channel instead of a matte you have to key out later. Verified on the live OAuth path 2026-08-21: 5/5 generations returned RGBA PNGs with all four corners at alpha 0 and 42-56% fully transparent pixels, including genuine PARTIAL alpha on glass and leaf veins. Saved as PNG; JPEG is refused because it cannot carry alpha.

Mechanics worth knowing: ChatGPT-session models pin the image tool to the gpt-image-2-codex variant, which rejects a FORCED background: "transparent" with a 400. ima2 therefore sends background: "auto" and puts the cutout intent in the prompt, which is what actually produces the alpha. You do not need to hand-write that suffix — --bg transparent adds it.

Use the solid-background-then-remove strategy only when you need a matte (chroma keying a video, compositing pipelines that expect green screen), or when a specific generation refuses to isolate the subject cleanly:

Generate on a pure solid background:

• Black (#000000) for reflective/metallic/glass subjects
• White (#ffffff) for dark/matte/opaque subjects
• Brand color when the target page background is known

State the exact hex and ban AI additions: "PURE SOLID BLACK background hex #000000. No checkerboard, no transparency pattern, no gradient, no floor plane, no shadow, no vignette." Use --mode direct.

ima2 gen "3D chrome splash on PURE SOLID BLACK background hex #000000. \
  No gradient, no floor, no shadow, no vignette." \
  --quality high --size 1024x1024 --mode direct -o splash.png

Remove background after generation:

• CSS mix-blend-mode: screen (black bg on light page)
• CSS mix-blend-mode: multiply (white bg on dark page)
• ima2 Canvas Mode background cleanup (export with alpha or matte)
• ima2 edit asset.png --prompt "remove the background, keep only the subject"
• Programmatic: sharp / ImageMagick / rembg

Anti-pattern: hand-writing "transparent background" into a prompt WITHOUT --bg transparent. Bare prompt wording sometimes yields a checkerboard pattern painted into an opaque image; the flag sends the real API parameter and the tuned suffix together. Always verify alpha rather than trusting the look of a preview: sharp(file).metadata() should report channels: 4, hasAlpha: true.

Raster-to-Vector (SVG)

ima2 gen always produces raster. When you need a REAL vector — a logo that scales, an icon for a design system, a brand-safe SVG asset — trace the raster afterwards. flat vector style in a prompt only produces a bitmap that LOOKS vector; it has no paths.

# 1. generate a cutout, 2. trace it
ima2 gen "a minimal geometric fox head logo mark, flat vector style" \
  --bg transparent --quality high --mode direct -o logo.png
ima2 vectorize logo.png -o logo.svg --json

Runs fully local — no server, no provider, no network. --json reports pathCount, bytes, and elapsedMs so an agent can check the result programmatically.

Presets: auto (default, best colour fidelity), flat (fewer colour layers), detailed, and mono (binary silhouette). Tune with --color-precision 1-8, --filter-speckle 0-128, --corner-threshold 0-180; pass a knob only when you actually want to override the preset, since any value replaces the tuned config.

Where it works and where it does not. Measured on a 1254x1254 keyed character asset: auto produced 722 paths at 585KB and is visually indistinguishable from the source. Cutouts, icons, logos, flat art, and sprite frames trace cleanly. Photographs and small text do NOT — text becomes illegible smudges and gradients fragment into hundreds of bands. Key or generate a clean cutout FIRST; tracing quality follows input flatness.

mono has no alpha channel, so a transparent cutout becomes a silhouette on a black field. Use it for stencils and line art, not for cutouts.

In the app, the same operation is Convert to SVG on an AssetGen tile or Assets library preview. Canvas Mode also exposes Trace to SVG (vector) in Export: it flattens the current canvas composition to PNG, saves that hidden canvas version, then opens the same preset and fine-tuning panel. Do not confuse it with SVG (embedded raster), which keeps the base image as bitmap data and vectorizes only Canvas annotations.

Korean Text in Images

When generating images with Korean text:

• Write the exact Korean string in quotes: "오늘의 추천", not "some Korean text"
• Describe the scene in English and keep only the visible Hangul string in Korean: A clean summer poster with the exact Korean headline "여름 축제". Practitioner testing found all-Korean prompts produced garbled Hangul while English prompts with a quoted Korean string rendered correctly (heuristic, not a guarantee)
• Start with short, label-like strings (a headline, a button) before attempting body copy; Hangul glyph complexity makes long dense text the most failure-prone case
• Specify font style explicitly: 고딕체 (Gothic/Sans-serif) or 명조체 (Myeongjo/Serif)
• Specify placement (top-center, bottom-left) and approximate size relative to the canvas
• For mixed Korean + English, specify which script appears where and in what hierarchy
• After generation, always inspect the result with view_image — garbled or substituted Hangul is common and must be caught before use
• For critical Korean text, generate 2-4 candidates (-n 4) and pick the cleanest render
• If a render is right except for the text, do a targeted ima2 edit pass that restates the exact string and changes only the text region; if spelling still will not stabilize after a couple of passes, stop retrying
• For legally or commercially exact Korean copy (packaging, UI, contracts), the reproducible production path is: generate the image with a reserved empty text area (no text in that region), then composite real type with an actual Korean font in an editor or code. Korean text failure is a cross-model limitation, not an ima2-specific one

Multi-Candidate Strategy

For important visual assets (hero images, key illustrations, brand materials), generate multiple candidates and select the best:

# 4 candidates from one prompt
ima2 gen "<detailed prompt>" -n 4 -d ./candidates --quality high

# Or multimode for structurally different directions
ima2 multimode "<detailed prompt>" --max-images 4 -d ./candidates

After generation, inspect every candidate with view_image before selecting. Do not blindly use the first result.

Prompt Iteration

• Start with one high-detail prompt. Inspect the result with view_image.
• On the next pass, make ONE targeted change and re-specify all constraints. Do not rewrite the entire prompt from scratch.
• Repeat invariants every iteration to prevent drift.
• This mirrors the official guidance: start from a clean baseline, iterate with small single-variable follow-ups instead of overloading one prompt, and when a detail drifts, restate it explicitly — never assume it persists.
• If the model consistently fails on a detail, try rephrasing, breaking the request into a base generation + ima2 edit pass, or switching --mode.

Frontend Asset Quick Recipes

Copy-paste starters for common frontend assets:

Hero image (landing page):

ima2 gen "Use case: product-mockup. Asset type: landing page hero. A premium wireless headphone floating at a slight angle against a soft warm-gray studio backdrop. Matte black finish with brushed aluminum accents. Soft three-point studio lighting, key light from upper-left. Shallow depth of field. Wide composition with generous negative space on the right for headline overlay. No text, no logos, no watermark." \
  --quality high --size 1536x1024 --mode direct -o hero.png

OG / social share image:

ima2 gen "Use case: ads-marketing. Asset type: social share card. Clean product flat-lay of a notebook, pen, and ceramic mug on a white marble desk. Overhead shot. Soft diffused daylight. Space in the upper third for title overlay. Warm neutral palette. No text, no logos, no watermark." \
  --quality high --size 1200x640 --mode direct -o og-image.png

App screenshot mockup background:

ima2 gen "Use case: stylized-concept. Asset type: hero background for device mockup. Soft abstract gradient from #f0f4f8 to #dbeafe with subtle geometric shapes at 5% opacity. Clean, modern, minimal. No objects, no patterns, no text." \
  --quality medium --size 1920x1088 --mode direct -o mockup-bg.png

Avatar / profile placeholder:

ima2 gen "Use case: stylized-concept. Asset type: user avatar. Friendly stylized portrait of a young professional, neutral expression, looking slightly left. Flat illustration style with subtle shadows. Solid #e5e7eb background. Circular crop safe. No text." \
  --quality medium --size 512x512 --mode direct -o avatar.png

Korean product hero:

ima2 gen "Use case: product-mockup. Asset type: Korean service landing hero. A modern smartphone at 15-degree tilt showing a clean fintech app UI. The screen displays a balance card with exact text \"잔액 1,234,500원\" in 고딕체, large centered. Soft gradient backdrop from #f8fafc to #e2e8f0. Studio lighting from upper-right. No other text, no logos, no watermark." \
  --quality high --size 1536x1024 --mode direct -o korean-hero.png

Game environment concept art:

ima2 gen "Use case: stylized-concept. Asset type: game environment concept art. A vast underground cavern with bioluminescent fungi on limestone walls. A narrow stone bridge crosses a dark chasm. Volumetric blue-green light from fungi clusters. Cinematic concept art style with industrial realism. Wide-angle, low camera, deep perspective. Mist rising from below. No characters, no text, no watermark." \
  --quality high --size 1792x1024 --mode direct -o cave-env.png

Reference / I2I Workflows

Reference generation:

ima2 gen "turn this into a clean product render" --ref input.png --quality high

Multimode reference workflow:

ima2 multimode "create four coherent variations" --ref input.png --max-images 4

Node-mode reference workflow:

ima2 node generate "continue this concept" --ref input.png

Image edit workflow:

ima2 edit input.png --prompt "make the object blue while preserving composition"

Do not use positional edit prompts. ima2 edit requires --prompt.

Structured Edit Brief

OpenAI's official edit pattern is "change only X" + "keep everything else the same" — an edit prompt does not need to re-describe the whole final image, but it must make the delta and the invariants explicit. Author every edit prompt as a brief:

Desired result: <one sentence describing the edited image's final state>
Change only: <the specific modification>
Preserve exactly: <named lock list: facial structure, pose, product
  silhouette, logo geometry, text spelling, framing, perspective, palette,
  lighting, shadows>
Do not add or remove: <protected elements>

"Keep everything else the same" alone is weak — name the fragile properties in the lock list, and repeat the same lock list on every iterative edit pass to prevent drift.

Annotated inputs. If the edit source or a reference image carries drawn markup (arrows, boxes, circled regions, sticky notes), the model tends to treat the markup as image content and reproduce it. Prefer sending the clean image plus text instructions derived from the markup. When the annotated image must be sent, state before and after the edit list that the markup is temporary editing instructions only — interpret it, apply the edits, then remove every trace of it from the output.

Removal edits. "Remove X" alone is weak. Pair the removal command with a positive description of what replaces it, then lock the rest: "Remove the sticky note. Show the continuous walnut desk surface where it was, matching the surrounding grain, lighting, and perspective — no residue, outline, or discoloration. Preserve every other object, the framing, and the color grading exactly." For stubborn removals, generate multiple candidates and re-edit only the residual region instead of enlarging the prompt.

Multi-Reference Rules

When passing multiple --ref images, label each reference by index and role inside the prompt, then state the relationships explicitly:

Image 1: base scene and composition.
Image 2: subject identity reference.
Image 3: style reference.

Place the subject from Image 2 into Image 1. Apply only Image 3's palette and
brushwork. Preserve Image 1's framing, background, perspective, and lighting.

• Put the most identity-critical reference (face, logo, product) first: documented GPT Image behavior preserves the first input with the richest texture and detail.
• When several faces must all stay recognizable, combine them into one composed reference image before generating instead of passing many separate portraits.
• For compositing, specify the source element, its destination and location, the preserved context, and harmonization: scale, perspective, lighting, shadows.

Parallel Generation

There is no --parallel flag. For multiple candidates from the same prompt, prefer one server-side batch request:

ima2 gen "four poster candidates" -n 4 -d ./out --quality high
ima2 multimode "four different poster directions" --max-images 4

For truly different prompts, independent CLI jobs can run concurrently against the same server. Capture request IDs with JSON output, then monitor or cancel:

ima2 gen "variation 1" --quality high --json
ima2 gen "variation 2" --quality high --json
ima2 ps --json
ima2 cancel <requestId>

Treat capabilities.limits.maxParallel as advisory client-side queue guidance only. It is not a guaranteed server-side semaphore.

Agent Mode (web UI only)

Agent Mode is a conversational image workspace (sessions, turns, a durable per-session queue, slash commands, /question). It is served at /api/agent/* and lives in the web UI — there is no ima2 agent CLI command. From the CLI, drive generation with ima2 gen, ima2 edit, ima2 multimode, and ima2 node generate instead.

Watching Jobs

Use JSON when another agent needs to reason about active work:

ima2 inflight ls --json
ima2 inflight ls --kind multimode --terminal --json

Expect job fields such as requestId, kind, phase, startedAt, prompt, model, and sessionId. Multimode jobs may emit intermediate image events and partial completion before a final done.

Prompt Import

Build a structured image prompt from a message or transcript:

ima2 prompt build --message "make this product prompt clearer" --json
ima2 prompt build --messages @conversation.json --json

Preview a local markdown/text prompt source before committing:

ima2 prompt import preview ./prompts.md --json

Import a JSON export body:

ima2 prompt import json ./prompts-export.json --folder __root__

Import a raw image into history:

ima2 history import ./local-image.png

Defaults

Inspect the running server defaults, including defaults.cli.image and defaults.cli.video in JSON:

ima2 defaults --json

Inspect local effective defaults without contacting a server:

ima2 defaults --local --json

Discover live model IDs and lane status before choosing a CLI target:

ima2 models
ima2 models --kind image --lane oauth --json
ima2 models --kind video --json

ima2 models --json has the stable shape {"ok":true,"kinds":{"image":[],"video":[]}}. It requires the server; an unreachable server returns SERVER_UNREACHABLE and exits 3.

Persist the server-side model defaults shared by GPT OAuth and API provider paths:

The built-in OAuth image default is gpt-5.6-luna; Grok image and video code defaults are grok-imagine-image-quality and grok-imagine-video respectively. For video, set grok-imagine-video-1.5 as your default and leave the base model for edit and extension, which are the only things 1.5 cannot do.

ima2 defaults set model gpt-5.5

Persist the fail-closed CLI image and video targets separately:

ima2 defaults set image oauth/gpt-5.6-luna
ima2 defaults set video grok/grok-imagine-video
ima2 defaults reset image
ima2 defaults reset video

Setting a CLI target validates the live catalog. Unknown models and lanes are rejected, and locked/disconnected/key-missing lanes cannot become defaults.

Persist the default reasoning policy:

ima2 defaults set reasoning high

Restart a running server after changing persisted defaults:

ima2 serve

Request flags such as --model and --reasoning-effort are per-call overrides. They do not change persistent defaults.

Capability Values

Use ima2 capabilities --json as the source of truth for:

• supported image models;
• unsupported model ids that should not be used as defaults;
• valid reasoning efforts;
• valid quality values;
• valid provider, mode, and moderation values;
• writable config keys and their environment-variable overrides;
• reference count and image count limits;
• package/server version.

Use only models from:

valid.imageModels.supported

Do not pick models from:

valid.imageModels.unsupported

Discover writable configuration keys:

ima2 config keys --json

Safety Notes

• Do not print API keys, OAuth tokens, config files, or .env values.
• Use ima2 capabilities --json before guessing model names.
• Use ima2 skill path when an agent needs the installed Markdown skill path.
• Use ima2 skill <name> refs to discover reference modules for front/uiux skills.
• Use ima2 skill <name> ref <refname> to load a specific reference module on demand.
• Use ima2 skill install --dir <path> to install skills to the agent's skill directory.
• Use ima2 inflight ls --json or ima2 ps --json to inspect active jobs.

Video Generation

Generate AI videos through a configured Grok or MCP lane. Grok OAuth requires a SuperGrok subscription; MCP lanes require their own connected subscription.

Quick Start

ima2 models --kind video
ima2 defaults set video grok/grok-imagine-video-1.5
ima2 video "a cat playing piano"                    # text-to-video, uses saved default
ima2 video "animate this" --ref photo.png           # image-to-video: photo is frame 1
ima2 video "same cat, on a beach" --ref cat.png --as-reference   # new scene, same subject
ima2 video "cinematic" --ref a.png --ref b.png      # 2+ refs are always references

Targets use --model <lane>/<model>; a bare ID is accepted only when it is unique across lanes. Generate-mode video also accepts an explicit --provider <lane>; run ima2 video --help for the lanes this build supports, since that list is derived from the provider registry rather than written here. --provider auto is removed.

Runway and Higgsfield are MCP lanes. They submit POST /api/mcp/generate (202) and the CLI waits on SSE until completion. MCP generation supports -n 1 only, and --ref values must be generated gallery filenames, not arbitrary local paths. Core-only planning/session flags are rejected with FLAG_NOT_SUPPORTED. ima2 gen and ima2 video also accept --character <element-id|name> on MCP lanes: the element must be a character element in the assets workspace with a provider binding for the selected lane (Runway: stateless refs + optional @tag). Fail-closed envelopes: CHARACTER_ELEMENT_NOT_FOUND, CHARACTER_ELEMENT_AMBIGUOUS, CHARACTER_BINDING_MISSING, CAPABILITY_MISMATCH (core lane or model without image_references), BINDING_NOT_READY, and server-side CHARACTER_ELEMENT_CONFLICT / CHARACTER_REFS_EXCEED_PROVIDER_CAP. Lane availability is runtime state, so it is not recorded here: a lane can be ready, missing a key, disconnected, or locked, and only a running server knows which. Read it with ima2 capabilities (per-lane status with the reason) or ima2 models --kind video (per-model rows). When ima2 capabilities reports source: local, no server answered and it carries no lane state at all — absence there means unknown, not unavailable.

ima2 upscale <generated-file> upscales through the MCP media-action pipeline: images take --scale-factor 2|4|8|16 (above 2 requires --flavor sublime), --flavor, --sharpen, --smart-grain, --ultra-detail; videos take no parameters. Multishot generation is POST /api/mcp/multishot (CLI surface planned). Video edit is the 2-step edit-video-preview → edit-video-submit media action; stage-1 returns a synchronous keyframe preview.

Model choice: reach for 1.5 first

grok-imagine-video-1.5 is the model to use unless you are editing or extending an existing video. It owns everything that makes a clip better; the base model owns two operations 1.5 refuses.

So: 1.5 for generating, base for editing and extending. Nothing needs both at once.

grok-imagine-video-1.5-preview is still accepted as a compatibility alias, but write grok-imagine-video-1.5 in anything new.

Prompt-only text-to-video on 1.5 is implemented as an internal white-canvas image-to-video anchor, because upstream 1.5 rejects raw T2V. That is an implementation detail — ask for text-to-video normally.

When a request does fall back to another model, the result carries requestedModel, effectiveModel, and modelFallback. Read effectiveModel before naming or reporting which model produced a clip. A request carrying --voice never falls back, because the base model cannot honor the voice and silently dropping it would return a clip missing what was asked for.

Modes (from --ref count, plus your choice at one reference)

One image is ambiguous and only the caller knows the intent, so it is a choice rather than a deduction. image-to-video locks that image as the opening frame and animates it. reference-to-video carries its subject, outfit, or location into a new scene without reproducing the shot. Two or more images can only be references.

Reference-to-video tops out at 720p; ask for 1080p there and the request is refused.

Parameters

Series Continuity (--topic)

--topic is legacy/best-effort series context. Prefer branch-local artifact continuity with ima2 video continue, Classic "Continue here", gallery video drag, or Node parent-video generation. Those flows use the previous generated video's last frame plus its stored revisedPrompt lineage.

Continuation is last-frame image-to-video, not true video-to-video. Only a single still frame carries over — motion, trajectory and camera movement are not preserved. Expect the next clip to start from that pose rather than to inherit momentum. Frames are pulled server-side with ffmpeg when the source is a generated file, falling back to in-browser canvas capture when ffmpeg is unavailable.

ima2 video "episode 1: morning routine" --topic "daily-vlog"
ima2 video "episode 2: commute" --model grok/grok-imagine-video --topic "daily-vlog"

Planning Layer

Prompts are NOT sent directly to the video model. A Grok planner rewrites your prompt with web search context for better results. The revisedPrompt in the response shows what was actually sent. Default planner model is grok-4.3 (configurable in settings UI).

Override the planner model per-request:

ima2 video "prompt" --model grok/grok-imagine-video --planner-model gpt-5.5
ima2 video "prompt" --model grok/grok-imagine-video --planner-model gpt-5.4

Grok 4.3 Prompt Surfaces

For video, the Grok 4.3 planner must produce one focused English prompt with: core subject, expected action/motion, camera/composition, environment/style, dialogue/audio intent, ending frame/continuity handoff, and constraints. If videoContinuity exists, the lineage is authoritative context: continue from the latest clip's final frame and final audio/dialogue state without restarting the scene. The planner also applies duration pacing: use the selected seconds as the full clip runtime, expand even short requests into a production-level sequence, and make the clip feel complete through composition, blocking, camera movement, motion rhythm, sound/dialogue timing, and an ending hold.

Active Video Prompt Requirement

Blank video prompts are blocked. Weak natural-language prompts are allowed, but agents should always write an active prompt that includes:

• shot design: opening frame composition, one motivated reveal or change, settling final frame — not a checklist of elements
• camera intent: choose the camera move that serves the scene (macro push-in for product, orbit for spatial VFX, handheld for documentary, crane for scale) — do not default to "slow dolly in"
• production choices: concrete material/texture, motivated lighting source, depth layers (foreground/mid/background), lens framing — instead of generic "cinematic" or "volumetric lighting"
• sound: music style, no music, room tone, or sound-effects-only
• dialogue: exact line in original language or explicit no-dialogue
• ending frame: final pose, camera state, last spoken words, and final sound cue — self-explanatory enough to serve as the first frame of a next clip
• duration pacing: beat structure scales with length — 1-4s gets one action, 5-7s gets setup/turn/hold, 8-10s gets two connected beats, 11-15s gets a three-beat arc

The planner is model-aware: it adjusts for grok-imagine-video (simpler, bolder composition at 480p) vs grok-imagine-video-1.5 (finer detail, 1080p textures). For 1.5 text-to-video, the server uses a white-canvas shim internally; the planner automatically writes a fresh-scene prompt without referencing a source image.

Anti-slop: the planner rejects generic prestige phrases ("AAA trailer", "senior VFX artist", "shot on RED"), filler lighting ("volumetric", "neon glow"), and unmotivated dark/moody defaults. Write what the camera actually sees.

Prerequisites

ima2 grok login     # authenticate (manual-paste flow)
ima2 grok status    # verify connection
ima2 serve          # server must be running

Output

SSE streaming events: planning → submitted → progress (0-100%) → done. The submitted and done payloads include requestedModel, effectiveModel, and modelFallback so agents can report which model actually produced a clip when a request falls back. A request carrying --voice never falls back. CLI --json prints video.requestedModel, video.effectiveModel, and video.modelFallback; use path/filename for local chaining.

Discover Valid Parameters

ima2 capabilities --json | jq '.valid.videoModels'

Advanced Workflows

Image-First Video (best quality)

Generate a high-quality still image first, then animate it. This produces better results than text-to-video alone because the video model has a concrete visual anchor.

Critical rule for i2v: Compose ALL characters and the environment together in ONE image. Do NOT use individual portrait refs for i2v — the video model needs a single composed scene to animate from.

Keyframe image provider rule (MANDATORY):

• Primary: GPT Image 2 (OpenAI, provider: oauth) with quality: high, maximum resolution matching the target video aspect ratio. For 16:9 video use 1792x1024. For 1:1 use 1024x1024. For 9:16 use 1024x1792.
• Fallback: Grok (provider: grok, model grok-imagine-image-quality). Only aspect ratio must match — resolution does not matter because i2v accepts any resolution source image and internally rescales.
• GPT Image 2 produces superior keyframes: better lighting coherence, character consistency, and fine detail that survives i2v animation. Always try GPT first.
• The i2v model internally rescales the source image to its native resolution regardless of input size, so there is no benefit to upscaling a Grok fallback image.

ref2v vs i2v decision:

# Multi-character scene: compose BOTH characters in one image first
# Primary: GPT Image 2 at high quality, max resolution, aspect ratio matching 16:9 video
ima2 gen "cinematic wide shot of Bruce Lee in yellow tracksuit facing Elon Musk in dark gi, underground fight arena, dramatic lighting, 16:9" --quality high --size 1792x1024 -o scene.png

# Fallback if GPT fails: Grok quality model, match aspect ratio only
# ima2 gen "same prompt" --provider grok --model grok-imagine-image-quality --size 1824x1024 -o scene.png

# Then animate from the composed scene
ima2 video "Bruce throws a rapid jeet kune do combination" --ref scene.png --duration 10 --resolution 720p --aspect-ratio 16:9

Multi-Shot Video (connected scenes)

Create a sequence of connected clips using --topic for narrative continuity. Each generation receives context from previous clips in the same topic.

# Scene 1: Establishing shot
ima2 video "wide establishing shot of a busy Tokyo street at night, neon signs" \
  --topic "tokyo-night" --duration 5

# Scene 2: Medium shot (planner sees Scene 1's revised prompt)
ima2 video "medium shot following a person walking through the crowd" \
  --topic "tokyo-night" --duration 5

# Scene 3: Close-up (planner sees Scenes 1+2)
ima2 video "close-up of rain drops on a neon sign reflection" \
  --topic "tokyo-night" --duration 5

The planner receives previous prompts from the same topic as continuity context. This is best-effort prompt guidance, not a guarantee that subjects, palette, or style will remain identical. For branch-local continuation, use ima2 video continue instead.

Storyboard-to-Video Chaining (9-panel storyboard → i2v loop)

The highest-quality video production workflow. Since Grok i2v accepts only one image input, pack the entire action sequence into a single 3×3 (9-panel) storyboard grid image. The i2v model reads the panels as a visual script and animates the progression.

Full workflow:

keyframe image (GPT high)
    → GPT i2i with reference → 9-panel storyboard grid
        → Grok i2v (reads panels, animates sequence)
            → extract last frame
                → GPT i2i with last frame → next 9-panel storyboard
                    → Grok i2v
                        → repeat

Step 1 — Opening keyframe (GPT Image 2, quality: high, max resolution matching target aspect ratio):

ima2 gen "cinematic wide shot of two fighters in a dojo, dramatic lighting" \
  --quality high --size 1792x1024 --storyboard

Fallback: Grok grok-imagine-image-quality, match aspect ratio only — resolution does not matter because i2v internally rescales.

Step 2 — 9-panel storyboard grid (GPT Image 2 with keyframe as reference):

# Use the keyframe as reference, prompt describes 9 sequential panels
ima2 gen "Using this scene as reference, create a 3x3 storyboard grid (9 panels, thin black borders) showing a 15-second action sequence. Panel 1 (0s): ... Panel 2 (2s): ... Panel 9 (15s): ... Maintain identical character designs across all panels." \
  --ref keyframe.png --quality high --size 1024x1024

9-panel storyboard rules:

• Grid layout: 3×3, thin black borders between panels
• Read order: left-to-right, top-to-bottom (panels 1-9)
• Panel 1 (top-left) MUST be solid black — this is a lead-in frame, not content. The i2v model starts from Panel 1's pixels; a black frame ensures the video begins with a clean fade-in instead of showing the grid. The 1-second black lead-in is auto-trimmed by the server.
• Panels 2-9 carry the action sequence (8 key moments with timestamps)
• Character designs MUST be identical across all panels
• Vary camera angle per panel for dynamic energy
• Each panel should look like a film still, not a sketch
• Do NOT add timestamp labels or text to panels — they burn into the video
• Square format (1024×1024) works best — i2v rescales internally

Step 3 — Animate storyboard via i2v:

ima2 video "This is a 9-panel storyboard. Animate the full sequence as one continuous 15-second clip following panels left-to-right, top-to-bottom. Panel 1: ... Panel 9: ... Sound: [describe music, SFX, dialogue]. Camera: [describe movement per beat]." \
  --ref storyboard.png --duration 15 --resolution 720p --model grok-imagine-video-1.5

i2v prompt rules for storyboard input:

• Explicitly state "This is a 9-panel storyboard" at the start
• Reference each panel by number with its action description
• Always include Sound/Music direction — never leave audio undefined
• Include Camera direction per beat (wide, close-up, tracking, handheld, slow-mo)
• Describe the end frame explicitly for continuation

Step 4 — Extract last frame and repeat:

# Extract last frame via ffmpeg
ffmpeg -sseof -0.1 -i clip.mp4 -frames:v 1 -q:v 2 -update 1 lastframe.jpg -y

# Generate next storyboard using last frame as reference
ima2 gen "Using this fight scene last frame as reference, create a 3x3 storyboard grid..." \
  --ref lastframe.jpg --quality high --size 1024x1024

# Animate next storyboard
ima2 video "This is a 9-panel storyboard..." --ref storyboard2.png --duration 15

Fallback: continueFromVideo — If a storyboard image triggers content moderation (common with intense action/fight scenes), fall back to video continue with a detailed text prompt instead:

ima2 video continue "detailed action description with sound and camera direction" \
  --video "$PREV_CLIP" --duration 15

Clip duration is flexible — use 15s for action-dense sequences with many beats, 10s for transitions, 5s for quick cuts. The 9-panel storyboard works best with 15s clips (each panel ≈ 1.5-2s of screen time).

Music and sound are MANDATORY in i2v prompts — describe the score (orchestral, percussion, taiko drums), sound effects (impacts, whooshes, crashes), dialogue lines, and audio transitions. "No music" or undefined audio produces flat, lifeless output.

Video Continuation (extend/sequel)

To continue from an existing video's last frame:

# Get the last generated video filename
LAST=$(ima2 ls -n 1 --json | jq -r '.items[0].filename')

# True extension keeps the original clip and appends new motion
ima2 video extend "the camera slowly pulls back revealing the full scene" --video "$LAST" --duration 6

# Branch-local sequel keeps revisedPrompt lineage and starts from the last frame
ima2 video continue "from the last frame, the camera slowly pulls back, no music, footsteps echo, end on a still wide shot" --video "$LAST"

Or in the UI: use "Continue here" on a video, drag a video from gallery/history to the prompt composer, or create a child from a video node. These flows attach the previous video's last frame and carry a branch-local videoContinuity lineage stack. The stack stores up to 4 revised prompts using keep-start-plus-latest-3: start clip is preserved, and the newest three clips stay in context.

ima2 video extend is xAI native extension: it returns original+extension as a combined artifact. ima2 video continue is ima2 branch continuation: it creates a new clip from the generated video's last frame and persists lineage metadata.

Marketing/Product Video

Generate a product showcase video from a product image:

# Step 1: Generate or provide product image
ima2 gen "clean product photo of wireless earbuds on white background" -o product.png

# Step 2: Create dynamic product video
ima2 video "sleek product reveal with rotating camera, premium feel, studio lighting" \
  --ref product.png --duration 10 --resolution 720p --aspect-ratio 16:9

Style-Consistent Series

For maintaining visual style across multiple videos (e.g., social media series):

# First video establishes the style
ima2 video "minimalist animation of a coffee cup, flat design, pastel colors" \
  --topic "coffee-series" --duration 5

# Subsequent videos inherit style via planner context
ima2 video "same style, now showing latte art being poured" \
  --topic "coffee-series" --duration 5

ima2 video "same style, steam rising from the cup" \
  --topic "coffee-series" --duration 5

Batch Generation (scripting)

#!/bin/bash
PROMPTS=("sunrise over ocean" "waves crashing" "seagulls flying" "sunset colors")
TOPIC="ocean-day"

for prompt in "${PROMPTS[@]}"; do
  ima2 video "$prompt" --topic "$TOPIC" --duration 5 --json >> results.jsonl
  sleep 2  # rate limiting
done

Limitations

• Max 15 seconds per clip (extend adds 2-10s more)
• Reference-to-video: 1-7 refs, up to 15 seconds, 720p ceiling. The 10-second clamp this section used to describe was ours, not xAI's, and was removed in v3.8.0 (issue #155)
• 1080p resolution is available for grok-imagine-video-1.5 prompt-only text-to-video via the white-canvas I2V shim, and for image-to-video with a single image/frame source
• Video edit/extend: grok-imagine-video only (1.5 is not supported)
• Video edit input: max 8.7 seconds
• Video extend input: 2-15 seconds; extension duration: 2-10 seconds

Provenance of the limits above

Do not re-litigate these from doc snippets alone. Two different kinds of fact are mixed in this section, and they age differently.

Owner-measured, not in any public doc — trust it, do not "correct" it away:

Live-probed against api.x.ai on 2026-08-20 (real requests, real responses):

The 1.5-vs-base split is not symmetric: 1.5 owns reference images, 1080p, and 15s; base owns video editing and extension. Neither model does both.

Video Editing (V2V)

Edit an existing video with a text prompt. This uses xAI's real video edit endpoint and saves the result as a generated video artifact.

# Get the local video

*Truncated - read the full file at https://github.com/lidge-jun/ima2-gen/blob/d2afe6b2aa7d006e2cd9765aa632714f96435db2/skills/ima2/SKILL.md.*