chromadna-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@chromadna-mcp Blend Portra 400 with Cinestill 400D and export a 3D LUT"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
ChromaDNA extracts reusable color settings from supported Lightroom / Camera Raw preset sources, blends compatible profiles, estimates photo-specific exposure and white-balance adjustments, renders approximate previews, and exports XMP presets and 3D LUTs.
Rendering scope: the preview renderer and .cube exporter are approximations. They do not reproduce every proprietary Adobe Camera Raw operation, camera profile, local adjustment, or RAW rendering stage. Validate generated presets and LUTs in the target application before production use.
Features
Preset mining from supported .xmp, embedded XMP in .dng, and legacy .lrtemplate sources.
Color DNA blending across supported sliders and tone curves.
Heuristic photo analysis and image-specific exposure / white-balance estimates.
Approximate Before / After preview rendering.
XMP preset compilation and 3D .cube LUT export.
Agent Skill and local FastMCP server.
Project-local setup for Claude Code, Cursor, Antigravity, and VS Code / GitHub Copilot.
Related MCP server: autochrome
Requirements
Python 3.10+
Node.js 22.20+ and npm for the optional multi-agent setup command.
uv for MCP server execution.
Git (the public repository can be cloned without a GitHub account).
Install the Python project
From the repository root:
python -m pip install -e .This installs the chromadna and chromadna-mcp commands with dependencies declared in pyproject.toml. The built wheel also bundles the default community catalog, so installed CLI/MCP commands can run outside the source checkout.
Quick start
# Mine your own local preset library into the catalog
chromadna mine ./raw_vault --out data/color_dna_catalog.json
# Inspect exact profile IDs before selecting one
chromadna list
# Blend two profiles and compile an XMP preset
chromadna blend --mix emulsion_kodak_portra_400:0.7 emulsion_cinestill_400d:0.3 --name "Hybrid Film" --out ./output_xmp/hybrid_film.xmp
# Render a preview and export a LUT from the same blend
chromadna blend --mix emulsion_kodak_portra_400:0.7 emulsion_cinestill_400d:0.3 --name "Hybrid Film" --out ./output_xmp/hybrid_film.xmp --preview-img ./sample.jpg --preview-out ./output_xmp/preview.jpg --lut ./output_lut/hybrid_film.cube
# Match a profile to a photo
chromadna match ./sample.jpg --dna emulsion_kodak_portra_400 --name "Matched Film" --preview ./output_xmp/matched_preview.jpg --out ./output_xmp/matched.xmp
# Export a standalone 33-cubed LUT
chromadna lut --dna emulsion_kodak_portra_400 --name "Warm Film" --size 33 --out ./output_lut/warm_film.cubeIf the package is not installed, use the repository launcher:
bash scripts/chromadna.sh listPowerShell:
.\scripts\chromadna.ps1 listMulti-agent setup
ChromaDNA is distributed from GitHub. It is not currently published to PyPI or npm. Clone the repository, then run:
node scripts/setup.mjs --allThe setup command installs the Agent Skill through the Vercel Skills CLI and configures native MCP entries for a wider set of coding agents, including OpenCode, Pi, Cline, Roo Code, Gemini CLI, Kimi Code CLI, Claude Desktop, and Windsurf. A few clients use global configuration files; existing entries are backed up and preserved. It does not publish packages or make the repository public.
Useful variants:
node scripts/setup.mjs --all --dry-run
node scripts/setup.mjs --skills
node scripts/setup.mjs --mcp --agent=cursorPrefer an AI to perform setup? Copy the single prompt in Install ChromaDNA with an AI Agent into any coding agent with terminal access. See Agent and Client Compatibility for the full matrix, DeepSeek Harness and Groq API adapters, and cloud-client limitations.
Community preset catalog
The default catalog contains 451 profiles imported from the pinned EmulsionPresets source revision. The imported records retain their source paths, revision, file hashes, and license information. ChromaDNA does not claim authorship of those preset recipes.
The catalog is not refreshed automatically. Maintainers can review and run the pinned importer manually; see Community Catalog Maintenance. Third-party notices and the import report live in data/sources/.
Releases
The repository is prepared for ChromaDNA 1.0.0. Use Actions → Release ChromaDNA → Run workflow and enter v1.0.0. The workflow resolves and commits uv.lock, runs the test and packaging checks, creates the matching Git tag, and publishes a GitHub Release with wheel and source archives. It does not publish to PyPI or npm.
Agent Skill
The canonical Agent Skill is in skills/chromadna/SKILL.md. The root SKILL.md remains as a compatibility copy for tools that discover a skill from the repository root.
MCP server
Run the local stdio server after installing the project:
chromadna-mcpAvailable tools:
list_color_dnainspect_color_dnaanalyze_photoblend_color_dnamatch_photo_to_dnarender_previewcompile_xmp_presetexport_cube_lut
The MCP process can only read files accessible to the machine and user account running it. Attaching an image in a cloud chat does not automatically expose its local filesystem path. The current server is not a hosted remote MCP endpoint.
Privacy and safety
Keep source photos, RAW files, and third-party preset binaries out of Git.
Keep
raw_vault/and generated output files local.Do not write generated presets beside RAW files using the same filename stem.
Do not claim compatibility until the preset or LUT has been tested in the target editor.
Do not expose the local stdio server to the public internet.
Contributing
Contributions are welcome. Start with CONTRIBUTING.md, read the Code of Conduct, and review Security before reporting a vulnerability.
License
MIT. See LICENSE.
Available Tools
8 toolsanalyze_photoA
Analyze an input photo's luminance, dynamic range, and RGB color cast, and recommend matching Color DNAs.
Args: image_path: Path to the target photo (.jpg, .png, .webp). top_k: Number of recommended Color DNA profiles to return (default: 3). catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: image_path (required), top_k, catalog_path. Returns: result
blend_color_dnaA
Mathematically blend 2 or more Color DNA profiles by weight (interpolating sliders & Tone Curves) and compile the hybrid style into an Adobe Lightroom .xmp preset.
Args: weights: Mapping of Color DNA ID to weight, e.g. {"film_emulation_001": 0.7, "cinematic_teal_orange_002": 0.3}. preset_name: Display name of the preset inside Lightroom. output_xmp_path: Destination path for the compiled .xmp file. slider_overrides: Optional dict of slider tweaks applied after blending (e.g., {"GrainAmount": 12}). preview_image_path: Optional source photo path to generate a Before/After JPEG preview. preview_output_path: Optional destination path for the Before/After preview JPEG. catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: weights (required), preset_name (required), output_xmp_path (required), catalog_path, +3 more. Returns: result
compile_xmp_presetB
Compile a Color DNA profile (with optional custom slider tweaks) into a valid Adobe Camera Raw .xmp file.
Args: dna_id: Color DNA ID from the catalog. preset_name: Preset display name inside Lightroom. output_xmp_path: Destination .xmp file path (must not collide with an existing RAW file stem). group_name: Preset folder/group name inside Lightroom (default: "ChromaDNA"). slider_overrides: Optional dict of Camera Raw slider adjustments. catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: dna_id (required), preset_name (required), output_xmp_path (required), group_name, +2 more. Returns: result
export_cube_lutA
Export a single Color DNA or a blended multi-DNA mix as a standard 33x33x33 .cube 3D LUT compatible with DaVinci Resolve, Adobe Premiere Pro, Final Cut Pro, and CapCut.
Args:
output_cube_path: Destination .cube file path.
dna_id: Single Color DNA ID (provide either dna_id or weights).
weights: Optional blend weights dict e.g. {"film_emulation_001": 0.7, "cinematic_teal_orange_002": 0.3}.
title: Title embedded inside the .cube file header.
size: 3D LUT cube resolution (default: 33, standard for video NLEs).
slider_overrides: Optional slider adjustments applied before baking the LUT.
catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: output_cube_path (required), size, title, dna_id, +3 more. Returns: result
inspect_color_dnaB
Inspect the full mathematical sliders and Tone Curves of a specific Color DNA profile.
Args: dna_id: The exact Color DNA ID (e.g., "film_emulation_001"). catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: dna_id (required), catalog_path. Returns: result
list_color_dnaA
List available Color DNA profiles from the catalog, optionally filtered by mood tag or category.
Args: tag: Optional mood tag filter (e.g., "matte", "warm-golden", "teal-shadows", "film-grain", "moody"). category: Optional category filter (e.g., "Film Emulation", "Cinematic Teal & Orange"). catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: tag, category, catalog_path. Returns: result
match_photo_to_dnaB
Auto-compensate Exposure and White Balance of a Color DNA profile to match a target photo's lighting,
then compile a tailored .xmp preset and optional Before/After comparison image.
Args: image_path: Path to the target photo (.jpg, .png, .webp). dna_id: Base Color DNA ID from the catalog. output_xmp_path: Destination path for the compiled .xmp file. preset_name: Display name inside Lightroom. slider_overrides: Optional manual slider adjustments applied after auto-matching. preview_output_path: Optional destination path for the Before/After JPEG comparison. catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: dna_id (required), image_path (required), output_xmp_path (required), preset_name, +3 more. Returns: result
render_previewA
Render a fast side-by-side Before/After JPEG preview applying a Color DNA onto an image without Lightroom.
Args: image_path: Source photo path (.jpg, .png, .webp). dna_id: Color DNA ID to simulate. output_preview_path: Destination JPEG path for the Before/After comparison. slider_overrides: Optional slider tweaks to preview before exporting XMP. catalog_path: Optional custom path to color_dna_catalog.json.
Inputs: dna_id (required), image_path (required), output_preview_path (required), catalog_path, +1 more. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 8 tools
Each tool has a mostly distinct role across discovery, inspection, analysis, blending, matching, previewing, preset compilation, and LUT export. There is some overlap because blend_color_dna, match_photo_to_dna, and compile_xmp_preset all produce .xmp files, and preview generation appears both as a standalone tool and as an option inside other tools.
All tool names use consistent snake_case with a clear verb_noun pattern such as list_color_dna, analyze_photo, compile_xmp_preset, and export_cube_lut. The convention is predictable throughout.
Eight tools is well-scoped for a color-profile pipeline covering catalog browsing, photo analysis, blending, preset compilation, preview rendering, and LUT export. Each tool has a distinct place in the workflow.
The surface covers the main lifecycle: discover, inspect, analyze, apply, blend, match, preview, and export. Minor gaps exist around catalog/profile management such as adding or updating custom Color DNA profiles, but core workflows are well covered.
Maintenance
Related MCP Connectors
Unofficial Color API MCP — wraps thecolorapi.com (free, no auth)
MCP server for StyleKit styles, tokens, recipes, lint and repair over stdio or HTTP.
MCP server for Clipkit — gives AI agents a video toolbox via the Clipkit schema.
A complete color workflow over MCP: mix, convert, harmonize, measure, place, and remember palettes.
Related MCP Servers
- AlicenseAqualityBmaintenanceA local, read-only MCP server that lets AI agents inspect and analyze photo libraries by scanning files, aggregating EXIF stats, finding duplicates, scoring blur, and generating cull reports without uploading any data.6MIT
- FlicenseNot gradedqualityDmaintenanceMCP server enabling AI agents to perform autonomous photographic color science and screenshot annotation with deterministic computer vision, film emulation, and real-time live preview.1-
- AlicenseNot gradedqualityCmaintenanceA local stdio MCP server that lets agents describe, render, inspect, and preview procedural 2D pixel-art skill effects and baked sprite animations entirely offline. It reads supplied local PNG/JSON projects and writes transparent sprite sheets, playback JSON, and image previews to a chosen local directory.MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to run cross-platform batch photo-processing workflows—including RAW decoding, adjustments, noise reduction, white balance, LOG restoration, EXIF tagging, deduplication, selection, HTML galleries, folder watching, and presets—through a unified versioned JSON contract. It also exposes MCP tools for processing, suggestions, autopilot, indexing, search, selection, HDR, face blurring, and more.MIT