Skip to main content
Glama
README.md
# Colour Memory

**The cultural colour intelligence layer for AI agents, designers, and brands.**

[![Smithery](https://img.shields.io/badge/Smithery-98%2F100-gold)](https://smithery.ai/servers/digby-oldridge/colour-memory-api)
[![colour-memory-api MCP server](https://glama.ai/mcp/servers/DigbyO/colour-memory-api/badges/score.svg)](https://glama.ai/mcp/servers/DigbyO/colour-memory-api)
[![UK Trademark](https://img.shields.io/badge/UK_TM-00004387450-navy)](https://colourmemory.com)

Tens of thousands of named colours across dozens of cultural archives. Every result provenance-labelled: documented fact, source type, and inferred hex always separated. Not generated. Not hallucinated.

---

## What it is

Every other colour API tells you what colours go together. Colour Memory tells you what they mean.

The difference is the archive. Tens of thousands of colours, each provenance-labelled with source type, cultural consequence data, and material context. No LLM names colours. No hallucination. Every result separates what is documented, what is inferred, and what is interpretive.

"Patna Opium Ball White. Bengal Opium Agency Annual Report 1833. The factory produced over 4,000 chests annually, graded by the whiteness of the outer coating."

That is a colour with a receipt. No other API produces that.

---

## Why it matters for AI

AI agents make colour decisions without cultural memory. They generate hex values with no understanding of what those colours carry historically, what they mean across markets, or whether they will land badly with a global audience.

Colour Memory gives agents a grounded colour intelligence layer. Every result includes archive source, cultural consequence, market risk, and confidence score. Hallucination risk is structurally eliminated because no LLM names colours -- the names come from the archive.

---

## The archives

Ancient Rome, Anglo-Saxon, Architecture, Austen, Australia, Brazil, Byzantine Empire, China, Dark History, Denmark, Dickens, EIC (East India Company), Finland, Football Strips, France, Georgian Pleasures, Germany, India, Islamic, Italy, Japan, Keats, Korea, Medieval Europe, Mexico, Military, Morocco, Netherlands, Nigeria, Norman Medieval, Norway, Ottoman Empire, Oxfordshire, Pigment, Pigment Historical, Plantagenet, Racing Silks, Roman Britain, Shakespeare, Space and Stars, Spain, Stuart Georgian, Sweden, Switzerland, Tingry, Tudor, USA, Viking Norse, Werner-Syme.

---

## Quick start

### Claude Desktop

```bash
claude mcp add --transport http colour-memory https://api.colourmemory.com/mcp
```

Or add to `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "colour-memory": {
      "type": "http",
      "url": "https://api.colourmemory.com/mcp",
      "headers": { "X-Api-Key": "YOUR_KEY" }
    }
  }
}
```

Get a free key (no card required): https://colourmemory.com/start

### Try these prompts in Claude

- What does the colour #D4A829 mean historically and culturally?
- Build a palette for a Japanese luxury brand avoiding colours with negative cultural associations.
- Is our brand colour #C0392B safe to use in a healthcare app targeting UK and Japanese markets?

### Any MCP client (ChatGPT, Cursor, Windsurf)

Add this URL: https://api.colourmemory.com/mcp


### REST API

    # Match a hex to the archive
    curl -X POST https://api.colourmemory.com/query/hex
      -H "X-Api-Key: YOUR_KEY"
      -H "Content-Type: application/json"
      -d '{"hex": "D4A829"}'

    # Query by concept or cultural brief
    curl -X POST https://api.colourmemory.com/query/conceptual
      -H "X-Api-Key: YOUR_KEY"
      -H "Content-Type: application/json"
      -d '{"query": "Ottoman luxury"}'

    # Brand audit
    curl -X POST https://api.colourmemory.com/brand/audit
      -H "X-Api-Key: YOUR_KEY"
      -H "Content-Type: application/json"
      -d '{"palette": ["#D4A829","#1A5C6E","#0F2D6B","#0A0A0B"]}'

Get your key at colourmemory.com

---

## Pricing

Explorer: Free, no card, 25 archive calls, 3 intelligence credits
Developer: Free, email required, 500 archive calls, 10 intelligence credits
Studio: £12/month, 2,000 archive calls, 50 intelligence credits
Pro: £39/month, 10,000 archive calls, 200 intelligence credits
Founder Enterprise: £299/month (20 slots only), 50,000 archive calls, 500 intelligence credits

Full details: https://colourmemory.com/pricing

---

## Architecture

- Runtime: FastAPI on Railway
- Search: CIEDE2000 nearest-neighbour with sentence-transformer semantic index
- Colour science: CIE Lab, LRV, chroma, three-illuminant simulation
- Auth: X-Api-Key header
- MCP endpoint: https://api.colourmemory.com/mcp
- Persistent storage: PostgreSQL on Railway
- Discovery: https://api.colourmemory.com/llms.txt · https://api.colourmemory.com/.well-known/agent.json

---

## Brand colours

Every Colour Memory brand colour was selected through the API itself.

- Song Gold Sycee #D4A829 -- China archive, Song dynasty
- Northanger Abbey Pool #1A5C6E -- Austen archive
- Saimaa Juhannus #0F2D6B -- Finland archive
- Velvet Black #0A0A0B -- Werner-Syme 1814
- Archive Parchment #F4EFE5 -- Architecture archive

---

## Legal

UK Trademark 00004387450, Classes 9, 35 and 42, 2026 PR Eye Ltd, Oxfordshire

colourmemory.com / hello@colourmemory.com

TDQS

A3.9/5.0

Scored across 66 tools

Disambiguation4/5

Most tools have distinct purposes with detailed descriptions, but the sheer number (66) means some conceptual overlap exists (e.g., multiple colour evaluation tools like colour_strategy, colour_verdict, colour_forensics). An agent would need to read carefully to avoid misselection.

Naming Consistency3/5

Naming follows a domain-prefix pattern (e.g., colour_, palette_, archive_), but conventions vary: some use verbs (accessibility_check), others nouns (colour_card), and there are different styles like query_conceptual (verb+adjective) and ui_states (prefix+plural noun). Overall, the system is readable but not strictly uniform.

Tool Count3/5

66 tools is high but justifiable given the extensive scope (colour query, palette generation, accessibility, brand, interior, ecommerce, archive management). It borders on overwhelming, but each tool addresses a specific need within the domain.

Completeness5/5

The tool set covers an unusually broad range of colour-related tasks: querying, mixing, comparison, naming, accessibility, cultural risk, brand audits, interior specification, image analysis, archive management, and more. No obvious gaps are apparent for the stated purpose.

Maintenance

ActivityActive
ResponsivenessSyncing