Skip to main content
Glama
Lulu-The-Narwhal

Dali MCP

by Lulu-The-Narwhal
OverviewSchemaRelated ServersScoreDiscussions
Python
Hybrid

Dali by Lulu

Score your prompt before you spend the credit.

Most AI generation failures are prompt failures. You can't tell the difference until after you've burned the token. Dali sits inside your agent and fixes that — before you generate.

You: "make a cinematic video of a woman walking in tokyo at night"

dali::score_prompt(prompt, "veo3")
→ 34/100  Grade: D
→ Missing: camera movement, lighting description, motion adverb
→ Verdict: High probability of a generic result. Enhance first.

dali::enhance_prompt(prompt, "veo3")
→ Gemini rewrites it using Veo 3's native language:

  "Cinematic. A woman in her 30s walks slowly through neon-lit Shinjuku
   at 3am, rain-wet streets reflecting pink and blue. Slow dolly push
   following behind her, low angle. Breath visible in cold air. Overcast
   amber light above, neon below. Melancholic, atmospheric. No text."

→ Score after: 84/100  Grade: A  ✓ Safe to generate.

Install

Hosted (recommended) — connect once, always-fresh guides, usage history:

# Claude Code
claude mcp add dali --url https://dali.getlulu.dev/mcp

# Cursor / Windsurf — add to MCP config:
{
  "mcpServers": {
    "dali": { "url": "https://dali.getlulu.dev/mcp" }
  }
}

Login with GitHub at dali.getlulu.dev — your history and scores sync automatically.

Self-hosted — local, no auth:

pip install dali-mcp
claude mcp add dali -- python -m dali.server

Related MCP server: Prompt Auto-Optimizer MCP

Tools

Tool

What it does

score_prompt(prompt, model)

Score 0–100 with grade, breakdown, what's missing, verdict

enhance_prompt(prompt, model)

Gemini rewrites the prompt in the model's native language

analyze_intent(prompt)

Parse dimensions: camera, motion, lighting, style, mood, gaps

my_story()

Your scoring history, model stats, grade distribution, insight

list_models()

All supported models with medium and strength

Supported models

Model

Medium

Core strength

veo3

Video

Cinematic camera language, photorealistic motion

higgsfield

Video

Physics-driven motion (cloth/hair/fluid), character consistency

kling

Video

Expressive character motion, facial performance

sora

Video

Temporal coherence, narrative sequences

midjourney

Image

Artistic style depth, community-proven patterns

flux

Image

Technical photography, camera/lens specificity, negative prompts

imagen

Image

Photorealism, lighting precision, photography brief language

Why model-specific?

Generic prompt optimizers don't know that Veo 3 needs camera movement more than anything else, that Midjourney ignores sentences and reads comma-separated descriptors, that Flux responds to camera body names like a photographer, that Higgsfield simulates physics so you describe materials in motion not motion abstractly, or that Kling reads expression language and generates facial performance from it.

Dali has a separate scoring model and a separate Gemini enhancement system prompt for each generator — because they speak different languages.

Model guides (MCP resources)

creative://guide/veo3         → Camera language + physics
creative://guide/higgsfield   → Motion + character consistency
creative://guide/midjourney   → Keywords + parameters
creative://guide/flux         → Technical photography brief
creative://guide/kling        → Expression + motion amplitude
creative://guide/sora         → Temporal coherence + sequences
creative://guide/imagen       → Photography brief language
creative://models             → All models overview

Contributing

Each model guide is in dali/data/guides/{model}.json. Found practitioner patterns that consistently work? Open a PR. The data format: prompt + model + quality_rating + notes. Every contribution improves the scorer.

MIT License · Built by Lulu · dali.getlulu.dev

This server cannot be installed

A
license - permissive license
-
quality - not tested
C
maintenance

How are these scores calculated?

Maintenance

–Maintainers
–Response time
–Release cycle
–Releases (12mo)
Commit activity

Resources

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Lulu-The-Narwhal/dali-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server