Omnigraph
Guides

Guides

Runnable graphs built with Omnigraph.

Each guide here corresponds to a real cookbook in the omnigraph-cookbooks repo , a self-contained schema, seed, queries, and omnigraph.yaml you can clone and run.

Industry Intelligence (SPIKE)

SPIKE-based industry intel graph. Signals, Patterns, Insights, KnowHow, and Elements for tracking a domain over time. Ships with a reference AI/ML seed (5 patterns, 15 signals, 111 nodes, 148 edges); generalizes to biotech, fintech, crypto, geopolitics, climate, or any industry where dated observations support or contradict recurring themes.

Pharma Intelligence

Pharma competitive intel. Extends SPIKE with two extra layers: an external pipeline (compounds, trials, deals, regulatory actions) and an internal context (programs, assumptions, decisions, open questions). Ships with a Viking Therapeutics GLP-1 reference seed (113 nodes, 200 edges, 15 node types). The graph surfaces which external signals contradict which internal assumptions before an upcoming decision.

Second Brain

Personal life ontology. People, places, notes, tasks, projects, habits, and media plus the provenance layer for capturing from any channel (email, WhatsApp, iMessage, voice, etc.). Ships with the fictional Alex Chen reference seed (121 nodes, 229 edges, 13 node types). The first non-SPIKE cookbook. Designed for individual life, not industry analysis.

Planned cookbooks

The cookbook repo also lists three more on the roadmap, not yet on this site:

  • company-context. Internal decisions, traces, actors, artifacts
  • biomed-research. Biotech and medical research tracking
  • competitor-intel. Competitor launches, pricing, positioning

Agentic skills

Two packaged skills live alongside the cookbooks for agents that want to set up or operate these graphs. See Agentic Skills.

On this page