What Ogham does
Features
Ogham is built around a simple loop: retrieve relevant experience, make a better decision, record the outcome. Everything below serves that loop.
Insights & episodes
Episodes are concrete past cases; insights are the generalised guidance distilled from them. Query the pattern, drill into the evidence.
Transparent scoring
Two independent axes — question-match relevance and trust/quality — shown separately, never multiplied into one opaque number.
Knowledge graph
An interactive graph of episodes, insights and tags, with a global view and a query-focused mode.
Hybrid retrieval
Semantic and lexical recall combined, with cross-lingual queries so the language of the question doesn’t limit the results.
MCP-native
Connects as a Model Context Protocol server — works with Claude Code and other MCP-capable agents out of the box.
Local-first & private
Runs on your own machine. Your decisions and data stay with you.
How a decision flows through Ogham
Before a consequential choice, your agent queries Ogham and weighs the returned insights. Once the decision is resolved, it records an episode. When the result becomes observable, it updates the outcome — and the next query is a little wiser.
See it in action
How it runs
Ogham is local-first: the service, your data and the web UI all live on your machine. Your agent talks to it over MCP.