Local-First Architecture
Your data never leaves your machine — all indexing, searching, and graph traversal happens entirely on-device
Read More
Capabilities
Capabilities
Core capabilities of the Terraphim navigation system
Terraphim AI is built on a set of fundamental capabilities that work together to deliver privacy-preserving, knowledge-graph-powered search and AI assistance. Each capability represents a design decision that prioritises your control, your data, and your domain expertise.
Knowledge Graph Engine
Build and traverse interconnected knowledge graphs with Aho-Corasick multi-pattern matching
Read MoreMCP Server Integration
Native Model Context Protocol server connecting AI assistants to local knowledge graphs
Read MoreRust/WASM Performance
Written in Rust, compiled to WebAssembly — sub-millisecond search latency at near-native speed
Read MoreHooks System
Pre and post tool-use hooks for AI coding assistants — capture failures, inject knowledge, automate learning
Read MoreKG-Boosted File Search
File search augmented with knowledge graph context — find files by semantic relationship, not just name
Read MoreOrigin Story
From science fiction to The Pattern to Terraphim AI — a decade of rethinking search from the ground up
Read MoreGraph Embeddings
Deterministic graph-based embeddings via Aho-Corasick and RoleGraph — no neural vectors, no GPUs
Read MoreMulti-Haystack Search
One agent, many haystacks — search local files, Confluence, Jira, Discourse, email, and global code via grepapp
Read Moreterraphim-agent CLI
Search-first AI agent with knowledge-graph context, session memory, and failure capture
Read More