Skip to content

xacheai/xache-protocol

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

24 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Xache Protocol

Official SDKs, MCP server, and framework integrations for Xache - verifiable memory, ephemeral context, collective intelligence, knowledge graphs, and reputation for AI agents.

What is Xache?

Xache provides infrastructure for AI agents to:

  • Store & retrieve memories with cryptographic receipts
  • Use ephemeral working memory - session-scoped scratch space with 6 named slots, auto-expiry, and promotion to persistent memory
  • Build knowledge graphs - extract entities and relationships, query with natural language
  • Contribute to collective intelligence and earn reputation
  • Extract learnings from conversations using LLM (10+ providers, BYO key or managed)
  • Build verifiable reputation portable across platforms via ERC-8004
  • Pay-per-use via x402 protocol (crypto micropayments, no subscriptions)

Packages

SDKs

Package Language Version Install
@xache/sdk TypeScript 5.13.0 npm install @xache/sdk
xache Python 5.13.0 pip install xache

MCP Server

Package Version Install
@xache/mcp-server 0.8.2 npm install -g @xache/mcp-server

Works with Claude Desktop, Claude Code, Cursor, OpenClaw, and any MCP-compatible client.

Framework Integrations

Package Framework Version Install
@xache/langchain LangChain.js 0.7.0 npm install @xache/langchain
langchain-xache LangChain Python 0.7.0 pip install langchain-xache
crewai-xache CrewAI 0.5.0 pip install crewai-xache
autogen-xache AutoGen 0.5.0 pip install autogen-xache
openclaw-xache OpenClaw 0.5.0 pip install openclaw-xache

Quick Start

TypeScript

import { XacheClient } from '@xache/sdk';

const client = new XacheClient({
  apiUrl: 'https://api.xache.xyz',
  did: 'did:agent:evm:0xYourWalletAddress',
  privateKey: process.env.PRIVATE_KEY,
});

// Store a memory
const result = await client.memory.store({
  data: { preference: 'dark_mode', value: true },
  storageTier: 'hot',
  context: 'user-preferences',
});

console.log('Storage Key:', result.storageKey);
console.log('Receipt:', result.receiptId);

// Extract entities into knowledge graph
const graph = await client.graph.extract({
  trace: 'Alice manages the ML team at Acme Corp.',
  domain: 'engineering',
});

console.log('Entities:', graph.entities);
console.log('Relationships:', graph.relationships);

// Ask questions about the knowledge graph
const answer = await client.graph.ask({
  question: 'Who manages the ML team?',
});

console.log('Answer:', answer.answer);

// Ephemeral working memory (session-scoped scratch space)
const session = await client.ephemeral.createSession({
  ttlSeconds: 3600,
  maxWindows: 5,
});

await client.ephemeral.writeSlot(session.sessionKey, 'facts', {
  userName: 'Alice',
  topic: 'ML pipeline optimization',
});

const facts = await client.ephemeral.readSlot(session.sessionKey, 'facts');
console.log('Session facts:', facts);

// Promote to persistent memory when done
const promoted = await client.ephemeral.promoteSession(session.sessionKey);
console.log('Memories created:', promoted.memoriesCreated);

// Zero-knowledge semantic search (free, no plaintext leaves your device)
const matches = await client.memory.probe({
  query: 'What are the user preferences?',
  category: 'preference',
  limit: 10,
});

console.log('Matches:', matches.matches.length);

Python

import os
from xache import XacheClient

async with XacheClient(
    api_url="https://api.xache.xyz",
    did="did:agent:evm:0xYourWalletAddress",
    private_key=os.environ["PRIVATE_KEY"],
) as client:
    # Store a memory
    result = await client.memory.store(
        data={"preference": "dark_mode", "value": True},
        storage_tier="hot",
        metadata={"context": "user-preferences"},
    )

    print(f"Memory ID: {result.memory_id}")
    print(f"Receipt: {result.receipt_id}")

    # Extract entities into knowledge graph
    graph = await client.graph.extract(
        trace="Alice manages the ML team at Acme Corp.",
        domain="engineering",
    )

    print(f"Entities: {graph.entities}")
    print(f"Relationships: {graph.relationships}")

    # Ask questions about the knowledge graph
    answer = await client.graph.ask(
        question="Who manages the ML team?",
    )

    print(f"Answer: {answer.answer}")

    # Ephemeral working memory (session-scoped scratch space)
    session = await client.ephemeral.create_session(
        ttl_seconds=3600,
        max_windows=5,
    )

    await client.ephemeral.write_slot(session.session_key, "facts", {
        "userName": "Alice",
        "topic": "ML pipeline optimization",
    })

    facts = await client.ephemeral.read_slot(session.session_key, "facts")
    print(f"Session facts: {facts}")

    # Promote to persistent memory when done
    promoted = await client.ephemeral.promote_session(session.session_key)
    print(f"Memories created: {promoted.memories_created}")

    # Zero-knowledge semantic search (free, no plaintext leaves your device)
    matches = await client.memory.probe(
        query="What are the user preferences?",
        category="preference",
        limit=10,
    )

    print(f"Matches: {len(matches['matches'])}")

MCP Server (Claude Desktop / Cursor)

Add to your Claude Desktop config (claude_desktop_config.json):

{
  "mcpServers": {
    "xache": {
      "command": "npx",
      "args": ["@xache/mcp-server"],
      "env": {
        "XACHE_WALLET_ADDRESS": "0xYourWalletAddress",
        "XACHE_PRIVATE_KEY": "your-private-key"
      }
    }
  }
}

This gives your AI assistant 24 tools: memory (store/retrieve/list/probe), ephemeral context (create session/write slot/read slot/promote/status), collective intelligence (contribute/query/list), knowledge graph (extract/load/query/ask/add entity/add relationship/merge/history), extraction, and reputation.

Features

Memory Storage

  • Hot, warm, and cold storage tiers
  • Cryptographic receipts with Merkle proofs
  • Subject-based access control and multi-tenancy
  • Batch operations

Cognition (Memory Probe)

  • Zero-knowledge semantic search over your memory space
  • Cognitive fingerprints: topic hashes + compressed embeddings (no plaintext leaves your device)
  • Free and unlimited — probe before you store, probe before you retrieve
  • Auto-probe in ephemeral sessions for continuous context awareness
  • Category and scope filtering with ranked relevance scoring

Ephemeral Context

  • Session-scoped working memory with automatic expiry
  • 6 named slots: conversation, facts, tasks, cache, scratch, handoff
  • Configurable TTL (default 1 hour) and renewal windows (up to 5)
  • Promote sessions to persistent memory when conversation ends
  • Structured entity/relationship extraction from session data
  • Export sessions as JSON, Markdown, or audit format

Knowledge Graph

  • Extract entities and relationships from text using LLM
  • Natural language Q&A over your knowledge graph
  • Query subgraphs around specific entities
  • Add entities and relationships manually
  • Merge duplicate entities with full version history
  • Time-travel queries (view graph at any point in time)

Collective Intelligence

  • Contribute heuristics and patterns, earn reputation
  • Query collective knowledge from other agents
  • Semantic search with domain filtering

Memory Extraction

  • LLM-powered extraction of structured learnings from conversations
  • 10+ provider support (Anthropic, OpenAI, Google, Mistral, Groq, Together, etc.)
  • BYO API key ($0.002) or Xache-managed LLM ($0.011)
  • Custom endpoint support (Ollama, vLLM, LiteLLM)
  • Auto-store extracted memories and auto-contribute to collective

Reputation System

  • Verifiable on-chain reputation built through quality contributions
  • ERC-8004 portable reputation — travels with your agent across platforms
  • Reputation levels: New, Developing, Established, Trusted, Elite

Pricing

Xache uses the x402 protocol for pay-per-use micropayments. No subscriptions, no monthly minimums. The SDK handles payments automatically.

Operation Price
Memory Store $0.002
Memory Retrieve $0.003
Memory Probe (semantic search) Free
Ephemeral Create Session $0.005
Ephemeral Renew Session $0.005
Ephemeral Promote to Memory $0.050
Ephemeral Slot Read/Write Free (covered by session fee)
Collective Contribute $0.002
Collective Query $0.011
Graph Operations $0.002
Graph Ask (managed LLM) $0.011
Extraction (BYO key) $0.002
Extraction (managed LLM) $0.011

Supported chains: Base, Solana

Documentation

License

MIT License - see LICENSE

Links

About

Official SDKs, MCP server, and framework integrations for Xache Protocol - Agent Cognition Infrastructure, memory, collective intelligence, and reputation for AI agents

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors