Social DNA

Canonical subject class definitions for AD4M applications.

Each DNA includes both TypeScript source (Ad4mModel classes) and SHACL JSON (for add_model via MCP). The JSON can be generated from the TypeScript using Ad4mModel.generateSHACL() — no AD4M executor needed.

Structure

<dna>/
  v1/                        ← Version directory
    Model.ts                 ← TypeScript source (Ad4mModel)
    Model.json               ← SHACL JSON (for add_model / MCP)
    index.ts                 ← Exports
  v2/                        ← Future versions
    ...

Schema Collections

Flux (flux/)

The 12 subject classes used by Flux:

Class	Description
Community	Top-level group with name, description, image, channels
Channel	Communication channel with views, participants, messages, tasks, posts
Message	Chat message with body, reactions, replies, threads
App	Application/view attached to a channel
Conversation	AI-detected conversation grouping
ConversationSubgroup	Sub-topic within a conversation
Topic	Named topic extracted from conversations
Embedding	Vector embedding for semantic search
SemanticRelationship	Expression-to-tag links with relevance
TaskBoard	Kanban board with ordered columns
TaskColumn	Column with ordered tasks
Task	Task with assignees and comments

Current version: v1

State of Affairs (soa/)

Knowledge representation for collective intelligence:

File	Description
StateOfAffair.ts/.json	Proposition with title, modality (belief/observation/intention/vision/plan/skill), confidence, status, tags, priority, source
predicates.ts/.json	9 relationship types: supports, contradicts, enables, requires, parent, refines, blocks, similar, same

Current version: v1

Memory (memory/)

AI agent persistent memory:

File	Description
MemoryEntry.ts/.json	Structured memory with content (markdown), timestamp, memoryType, importance (1-10), tags, author DID, shareable flag

Current version: v1

Usage

For AI agents (via MCP)

add_model(perspective_id, "MemoryEntry", <contents of memory/v1/MemoryEntry.json>)

For TypeScript apps

import { MemoryEntry } from '@coasys/social-dna/memory/v1';
import { StateOfAffair, SoA } from '@coasys/social-dna/soa/v1';

// Use in a perspective
const entries = await MemoryEntry.findAll(perspective);

// SoA relationships
await perspective.addLink({
  source: soaA.id,
  predicate: SoA.REL_SUPPORTS,
  target: soaB.id
});

Generating SHACL JSON from TypeScript

import { MemoryEntry } from './memory/v1/MemoryEntry';

const { shape } = (MemoryEntry as any).generateSHACL();
fs.writeFileSync('memory/v1/MemoryEntry.json', JSON.stringify(shape, null, 2));

Versioning

Each DNA collection uses directory-based versioning (v1/, v2/, etc.). When a schema changes in a breaking way, create a new version directory. Non-breaking additions can be made within the same version.

Contributing

1. Write the TypeScript model using @Model / @Property / @Optional / @HasMany from @coasys/ad4m
2. Generate the SHACL JSON (or write it manually — they're equivalent)
3. Place both in the appropriate <dna>/v<N>/ directory
4. Update this README
5. Submit a PR

License

MIT