Dali scores and upgrades image and video prompts before generation, preventing wasted credits and improving output quality.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install dali
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install dali using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Dali Creative Intelligence?
Dali is an Openclaw Skills preflight layer for image and video generation: it scores every prompt before a credit is spent, highlights what is missing, and rewrites weak prompts with model-aware guidance.
By combining MCP-based scoring with model-specific rules for tools like Veo, Seedance, Kling, Runway, Flux, Midjourney, Ideogram, and Firefly, Dali helps teams turn rough ideas into production-ready prompts, reduce wasted generations, and ship more consistent creative output.
Dali Creative Intelligence Use Cases
- When a user wants to generate, create, render, or animate an image or video with a model such as Veo, Sora, Kling, Runway, Seedance, Midjourney, Flux, Ideogram, or Firefly.
- When a raw prompt is pasted and needs quality control before any generation credit is spent.
- When the target model is unclear and the agent needs to infer the best option from context.
- When prompts are generic, under-specified, or likely to produce stock-looking or low-fidelity output.
- When a team wants Openclaw Skills to enforce a consistent prompt-quality gate across creative workflows.
- When the user wants model-specific benchmarks, creative pattern guidance, or prompt history insights.
How Dali Creative Intelligence Works
- Detects whether the request is about creating an image or video and infers the likely model from context when one is not specified.
- Calls
score_prompt(prompt, model)immediately and returns a concise human-readable grade, score, missing elements, and verdict. - Applies the grade policy: A/B prompts are ready to generate, C prompts can be enhanced, and D/F prompts are enhanced automatically.
- When enhancement is needed, calls
enhance_prompt(prompt, model)to retrieve a rewrite brief containing model language rules, structure guidance, priorities, and missing elements. - Writes the improved prompt from that brief, then re-calls
score_prompton the rewritten version to verify the score delta. - Uses the highest-scoring prompt for generation and optionally calls community tools for benchmarks, pattern discovery, model support, or prompt history.
Dali Creative Intelligence Setup
- No local installation or API key is required for scoring.
- Point your MCP-enabled agent to the Dali endpoint.
export DALI_MCP_ENDPOINT="https://dali.getlulu.dev/mcp"
# Register this endpoint in your MCP client or agent configuration
- Configure your workflow so Dali runs before any image or video generator.
- If a prompt scores low, have the agent enhance it and re-score the rewritten version before generating.
Dali Creative Intelligence Data Schema & Taxonomy
| Artifact | Shape | Purpose |
|---|---|---|
| Prompt input | string |
Raw user prompt, pasted draft, or generation request. |
| Model selector | string |
Explicit or inferred target model such as veo3, seedance, kling, runway, wan, minimax, higgsfield, flux, midjourney, ideogram, or firefly. |
| Score result | { grade, score, missing[] } |
Immediate prompt-quality assessment before generation. |
| Enhancement brief | { native_language_rules, structure_template, priority_fixes, length_target, what_to_add } |
Guidance used to author the improved prompt. |
| Enhanced prompt | string |
The final rewritten prompt that should be sent to the generator. |
| Verification result | { score_delta, re_score } |
Confirms that the rewritten prompt improved materially. |
| Community tools output | { benchmark, patterns, history, models } |
Optional supporting context for deeper optimization. |
- Grade taxonomy:
A= 90+,B= 70–89,C= 50–69,D/F= below 50. - Output policy: human-readable guidance only; the skill does not dump raw JSON to the user.
- File behavior: no files are written locally by default; outputs are returned as transient assistant text and code blocks.
- Metadata focus: model-specific prompt quality, missing creative elements, and before/after score deltas.
Dali Creative Intelligence Advanced Features
- Automatic intent detection for any image or video generation request.
- Model inference when the user does not specify a target model.
- Grade-based branching that auto-enhances D/F prompts and offers enhancement for C prompts.
- Model-specific rewrite briefs instead of generic prompt polishing.
- Re-scoring loop that proves the improvement before generation proceeds.
- Community tools such as
community_benchmark(prompt, model),creative_patterns(model),my_story(), andlist_models(). - Zero-friction MCP access on the free tier with no API key required for scoring.
- Human-readable output designed for fast decision-making in Openclaw Skills workflows.
SKILL.md
Loading
Related Openclaw Skills
OpenClaw VPS Security Preflight
A read-only VPS security preflight that validates OpenClaw exposure, SSH/firewall posture, and deployment readiness on authorized Linux hosts.
Last Mover Advantage
A strategy skill that tests whether a market rewards being first, or rewards the durable last mover that wins the endgame.
Definite Optimism
An analytical framework that audits organizational behaviors and redirects teams from optionality to concrete, falsifiable long-term strategies.
Faceless Reels Generator
An AI workflow that turns a topic into a complete faceless short-form video package for Reels, TikTok, and Shorts.
savemaintenance
savemaintenance is an Openclaw Skills maintenance pipeline that backs up, reconciles, rebuilds, and audits the three-tier memory system in one deterministic run.
Save — Conversation Memory Index
Save is an Openclaw Skills memory system that turns live conversations into durable, searchable local knowledge.
Featured*
