Context SpineContextSpine
Back to Home

Context Spine Documentation

Complete guide to the Context Spine platform - your organizational intelligence system for tracking events, entities, and insights across all your data sources.

Dashboard

Workspace overview and recent activity

Radar

Live event visualization

Ask

Query data with natural language

API Reference

REST API documentation

Introduction

Context Spine is an organizational intelligence platform that helps you track what's happening across your data sources, understand why it matters, and see how things are changing over time. It transforms scattered data into unified, actionable intelligence.

Key Concepts

  • Events - Immutable records of things that happened, with bitemporal timestamps and value classification (hot/warm/cold)
  • Entities - Stable identifiers that unify people, organizations, and objects across different systems
  • Assertions - Derived insights with confidence scores and lifecycle states
  • Nerves - Data source connectors (GitHub, Hacker News, Reddit, etc.)
  • Cerebrums - Analysis modules that generate assertions from events

Core Architecture

Context Spine follows a three-tier data processing architecture:

1

Nerves - Data Collection

Nerves are data source connectors that ingest events. They include API integrations (Hacker News, GitHub, Stack Exchange), browser scrapers (Reddit, LinkedIn), and custom webhooks.

2

Cortexes - Orchestration

Cortexes are intelligent orchestration layers that direct nerves and process events. They manage platform-specific logic (like Reddit Cortex) and coordinate data flows.

3

Cerebrums - Analysis

Cerebrums are analysis modules that derive insights from events. Examples include Whitespace Detector (finds market gaps) and Brand Monitor (tracks brand mentions and narratives).

Dashboard

The Dashboard provides a high-level overview of your workspace activity with key statistics and recent events.

Key Metrics

Total Events

All events in your workspace

Hot Events

High-priority events in the last hour

Data Sources

Active nerves currently running

Last Event

Most recent event timestamp

The dashboard also shows events by source distribution and a table of recent events that auto-refreshes every 30 seconds.

Radar

The Radar provides a live visualization of events flowing into your workspace. Events appear as blips on a circular radar display, moving from the outer edge toward the center as they age.

Features

  • Time Window: View events from 15 minutes to 24 hours
  • Color Modes: Color by value class (hot/warm/cold) or by source
  • Source Filtering: Toggle which data sources to display
  • Health Panel: View nerve status and system health
  • Voice Announcements: Optional TTS for hot events

Tip: Click any blip on the radar to navigate directly to that event's detail page.

Ask (Natural Language Querying)

The Ask feature lets you query and analyze your data using natural language. It uses a local LLM (Ollama or LM Studio) to understand your questions, generate data pipelines, and provide insights.

How It Works

1

Intent Analysis: The LLM understands your question

2

Schema Discovery: Identifies relevant data tables

3

Pipeline Generation: Creates a data query pipeline

4

Execution & Analysis: Runs the pipeline and summarizes results

Example Questions

"What were the top 5 starred GitHub repos today?"
"What's the average HackerNews score this week?"
"Show me recent high-priority events"
"What topics are trending on Reddit?"

Diver

Diver lets you deep-dive into specific events and track their evolution over time across different platforms.

Supported Platforms

Reddit

Track threads, comment growth, engagement patterns

Hacker News

Explore stories, score-to-comment ratios, trends

GitHub

Track repos, star patterns, popular projects

Events

Events are the foundation of Context Spine. Every piece of information is represented as an immutable event in the Context Ledger.

Event Properties

{
  "id": "evt_abc123",
  "event_type": "hackernews:story:scored",
  "source_system": "hackernews",
  "event_time": "2024-01-15T10:30:00Z",  // When it happened
  "record_time": "2024-01-15T10:30:05Z", // When we learned about it
  "value_class": "hot",                   // hot | warm | cold
  "authority_class": "source_of_truth",   // source_of_truth | observer | inference
  "payload": {
    "title": "Show HN: My new project",
    "score": 142,
    "url": "https://example.com"
  },
  "linked_entities": [
    { "entity_type": "person", "display_name": "user123" }
  ]
}

Value Classes

Hot

High-priority, requires attention

Warm

Notable, worth monitoring

Cold

Background data, routine

Bitemporal Model

Every event has two timestamps:

event_time

When the event actually occurred

record_time

When Context Spine learned about it

Entities

Entities are stable identifiers that unify references to people, organizations, repositories, and other objects across different systems.

Entity Types

personorganizationrepositoryissuepull_requestdealcontactslack_channel

Each entity tracks event count, last seen timestamp, activity sparklines, and external IDs from different source systems.

Assertions

Assertions are derived insights that Context Spine computes from events. They represent claims about the world with confidence scores and lifecycle states.

Assertion Lifecycle

activesupersededretracteddisputedexpired

Score Breakdown

Assertions include detailed score breakdowns:

Problem Intensity

Strength of the identified problem

Supply Gap

Lack of existing solutions

Novelty

How new/unique the opportunity is

Coherence

Consistency of supporting evidence

Embeddings

Embeddings are vector representations of event content that enable semantic search. When events are ingested, their text content is embedded into high-dimensional vectors.

Key Stats

  • Total embeddings in the system
  • Vector dimension (features per embedding)
  • Embeddings created in the last 24 hours
  • Breakdown by source system

Brand Intelligence

The Brands module monitors brand mentions and narratives across social platforms, identifying risk signals and tracking sentiment.

Explorer

Browse monitored brands with narrative counts, mention counts, and tags.

Risk Signals

View detected risk signals with severity levels (high/medium/low).

Sources

Configure which platforms to monitor for brand mentions.

Settings

Add new brands with aliases and search terms.

Nerves (Data Sources)

Nerves are data source connectors that ingest events into Context Spine. Each nerve has health status, event counts, and error tracking.

Available Nerves

Hacker News
GitHub Public
Stack Exchange
Wikipedia
arXiv
Dev.to
Lobsters
XKCD
NASA
Finance
Crypto
Weather
Earthquakes

Browser Scraper Plugins

Some platforms are scraped via a browser extension:

Reddit

Collects posts, comments, engagement metrics

LinkedIn

Collects posts, comments, profile data

Cortexes (Orchestration)

Cortexes are intelligent orchestration layers that direct nerves and process events. They handle platform-specific logic and coordinate data flows.

Status Information

  • Online/Offline: Current operational status
  • Version: Currently deployed version
  • Uptime: Time since last restart
  • Events Received: Total events processed
  • Commands Sent: Control commands issued

Cerebrums (Analysis)

Cerebrums are analysis modules that generate assertions from events. They can be enabled/disabled and configured per workspace.

Available Cerebrums

Whitespace Detector

Analyzes problem signals and supply signals to identify market gaps and opportunities where user needs are unmet by existing solutions.

min_cluster_size: 5min_confidence: 0.5window_days: 30

Brand Monitor

Detects brand mentions across social platforms, clusters them into narrative themes, and identifies risk signals like volume spikes.

min_cluster_size: 3similarity_threshold: 0.7window_days: 7

System Errors

The Errors page provides visibility into nerve errors and system health. Errors are aggregated from Prometheus metrics with configurable time windows.

Features

  • Filter by time window (15 min to 24 hours)
  • Filter by nerve or error type
  • Auto-refresh every 10 seconds
  • Nerve summary cards with error counts
  • Expandable error context details

Settings

Configure your workspace settings including general options, assertion decay policies, and API access.

General

  • Workspace name
  • Description
  • Timezone
  • Data retention period

Assertions

  • Default TTL (days)
  • Require reaffirmation
  • Decay policy settings

API

  • API endpoints
  • Authentication token
  • Documentation link

API Reference

Context Spine provides a REST API for programmatic access to all features.

Authentication

All API requests require a Bearer token in the Authorization header:

Authorization: Bearer YOUR_API_TOKEN

Key Endpoints

POST/events/ingestIngest events
GET/eventsList events with filtering
GET/events/searchFull-text search events
GET/assertionsList assertions
GET/entitiesList entities
POST/askNatural language query
GET/nerves/statusNerve health status

Quick Start

Get started with Context Spine by ingesting your first event.

Ingest an Event

curl -X POST http://localhost:3060/events/ingest \
  -H "Authorization: Bearer YOUR_API_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "workspace_id": "YOUR_WORKSPACE_ID",
    "events": [{
      "event_type": "custom:user:action",
      "event_time": "2024-01-15T10:30:00Z",
      "source_system": "my-app",
      "value_class": "warm",
      "payload": {
        "user_id": "user123",
        "action": "signup"
      }
    }]
  }'

Need help? Contact our team for enterprise support.

Contact Sales