Welcome to Crew10X Documentation

Single-task executors optimized for speed and cost efficiency. Ideal for repetitive, well-defined operations that require minimal reasoning.

Domain-specific experts with advanced reasoning capabilities and persistent vector memory. They develop expertise over time through continuous learning.

Vector Memory

High-dimensional embedding store for semantic search and similarity matching. Every piece of information an agent encounters is encoded into dense vector representations, enabling fuzzy recall across sessions.

memory = agent.memory.vector

memory.store(
    content="Q3 revenue exceeded projections by 14%",
    metadata={"domain": "finance", "quarter": "Q3-2024"}
)

results = memory.search("revenue performance", top_k=5)

Knowledge Graph

A structured relationship network that maps entities, concepts, and their connections. Enables agents to perform multi-hop reasoning and discover non-obvious relationships in their knowledge base.

graph = agent.memory.knowledge_graph

graph.add_entity("NVIDIA", type="company")
graph.add_entity("H100", type="product")
graph.add_relation("NVIDIA", "manufactures", "H100")

paths = graph.query("NVIDIA", hops=3)

Working Memory

Ephemeral, session-scoped memory that holds the agent's current reasoning state, active hypotheses, and intermediate computations. Automatically garbage-collected when reasoning cycles complete.

working = agent.memory.working

working.set("current_hypothesis", "Supply chain disruption is primary driver")
working.set("confidence", 0.73)

state = working.snapshot()

import cog_os
from cog_os.agents import Worker, Specialist, Executive
from cog_os.memory import VectorStore, KnowledgeGraph

client = cog_os.Client(api_key="sk-crew10x-...")

swarm = client.swarms.create(
    name="research-team",
    agents=[
        Specialist(role="data-analyst"),
        Specialist(role="report-writer"),
        Worker(role="data-collector", count=3),
    ],
    executive=Executive(role="project-lead")
)

result = swarm.execute("Compile competitive analysis for Q1 board meeting")

Requires Python 3.10+. Install via pip install cog-os

import { Crew10X, Specialist, Worker } from '@crew10x/sdk';

const client = new Crew10X({ apiKey: 'sk-crew10x-...' });

const agent = await client.agents.create({
  type: 'specialist',
  name: 'code-reviewer',
  consciousness: true,
  memory: { vectorStore: '10gb', knowledgeGraph: true }
});

const result = await agent.deploy({
  objective: 'Review all PRs in the last 24 hours'
});

Requires Node.js 18+. Install via npm install @crew10x/sdk

REST API

Base URL: https://api.crew10x.com/v1

WebSocket API

Real-time streaming endpoint: wss://stream.crew10x.com/v1

const ws = new WebSocket('wss://stream.crew10x.com/v1');

ws.onopen = () => {
  ws.send(JSON.stringify({
    type: 'subscribe',
    agent_id: 'agent_abc123',
    events: ['reasoning', 'memory', 'execution']
  }));
};

ws.onmessage = (event) => {
  console.log(JSON.parse(event.data));
};

Authentication

All requests must include your API key in the Authorization header.

$ curl -X POST https://api.crew10x.com/v1/agents \
  -H "Authorization: Bearer sk-crew10x-your-key" \
  -H "Content-Type: application/json" \
  -d '{"type": "specialist", "name": "analyst"}'

Generate API keys from your dashboard. Keys are scoped to organization-level access.